A radical request - replace the default MP3 decoder

[Roj]

No, I'm not going to suggest MAD - licensing issues with the GPL would likely make that quite untenable.

I am going to suggest BASS, which is the heart of several rival players and to my ears every bit as good as MAD. It is a 32-bit decoder and can be paired with dithering and soundshaping a la MAD and is even tweakable for seamless gapless MP3 playback. The author of XMPlay created and licenses it:

www.un4seen.com (http://www.un4seen.com)

It is decidedly better than the outdated and clunky XAudio decoder QCD currently uses from an audio quality perspective and with this engine the player would completely obviate any need for plugins such as MAD and <shudder> MPG123.

Paul?

[jkrzok]

I'm impressed by XMPlays sound quality and would wecome Roj's suggestion.

Is it available as a Winamp plugin?

It's a shame Roj couldn't 4see that un4seen.com would be down at about the time he made his post.

[hedge]

Yep I agree entirely that the default mp3 decoder needs to be replaced. And I gotta say that xmplay does sound good, so it would seem to be a very good choice.
Tell me one thing, does it actually manage to calculate song lengths correctly all the time, unlike xaudio, which is a mixed bag in that department to say the least.

[Toe]

No, I'm not going to suggest MAD - licensing issues with the GPL would likely make that quite untenable.
Maybe it's time to GPL QCD? ;)

(Hey, I wasn't expecting Shareaza to be GPL'd a couple months back, but that's just what happened...)

[Roj]

Maybe it's time to GPL QCD? ;)

(Hey, I wasn't expecting Shareaza to be GPL'd a couple months back, but that's just what happened...)
I don't think Paul can since the Gracenote stuff is commercial. As to calculating song times correctly (hedge's statement), BASS doesn't appear to have the problem. It's cheap too - 100 Euros for an unlimited license. If it were my choice, I wouldn't do a plugin - I'd just replace XAudio and be done with it. Of course that is likely considerably more problematic than it sounds... :)

[brian]

Interesting suggestion, Roj. You seem to be coming round to my view that the only person who can solve the QCD plugins problem is Paul Quinn...

I entirely agree with you about the quality of the BASS decoder. But I don't quite see the value of importing it into QCD. Why not just use XMPlay? QCD has better skins and a wider array of preferences, but that's about it.

I've taken to using CoolPlayer when I want the MAD sound. It will play any mp3 stream I throw at it, whereas the QCD plugin (Shao Hao's) chokes on at least 50% of streams.

We await developments! It wouldn't surprise me if Paul is already grappling with this kind of issue. Of course there may be other explanations for the lengthening silence...

[Roj]

Interesting suggestion, Roj. You seem to be coming round to my view that the only person who can solve the QCD plugins problem is Paul Quinn...Not at all - I firmly believe that developers can and should build for the player. After all, they do for Winamp and even fubar. Do note also that XMPLay relies completely on Winamp plugins and has no SDK for that facility at this time.

I entirely agree with you about the quality of the BASS decoder. But I don't quite see the value of importing it into QCD. Why not just use XMPlay? QCD has better skins and a wider array of preferences, but that's about it.Those better skins and wider array of options go a VERY long way. So do an existing community dedicated to quality, continual large strides in development and versatility. For example, XMPlay can't even play CDs natively.

I've taken to using CoolPlayer when I want the MAD sound. It will play any mp3 stream I throw at it, whereas the QCD plugin (Shao Hao's) chokes on at least 50% of streams.I was once an adamant CoolPlayer supporter but I can no longer tolerate that player. When Marcus joined the development team, it was the worst possible thing that could have happened to that player. He instituted sweeping changes, didn't test them thoroughly, developed sporadically and in the end left the project high and dry in a decidely chaotic state. That playlist is a retroactive abortion.

I've already commented on the build quality of Shaohao's plugins as politely as I could elsewhere.

We await developments! It wouldn't surprise me if Paul is already grappling with this kind of issue. Of course there may be other explanations for the lengthening silence...Fatherhood is VERY time consuming. I should know - I've got three kids to contend with. :)

[brian]

Roj - well answered. For the time being I've given up on the idea of having one player that does everything brilliantly. I keep four players on my system (QCD, XMPlay, CoolPlayer, WMP) and decide which to use according to what I want to listen to at any given moment. QCD is easily best for skins and preferences, but sound quality and stream-connectivity are my main priorities.

Which version of CoolPlayer would you recommend as the last good build?

I'll shut up now as I'm off topic.

[Roj]

Roj - well answered. For the time being I've given up on the idea of having one player that does everything brilliantly. I keep four players on my system (QCD, XMPlay, CoolPlayer, WMP) and decide which to use according to what I want to listen to at any given moment. QCD is easily best for skins and preferences, but sound quality and stream-connectivity are my main priorities.

Which version of CoolPlayer would you recommend as the last good build?

I'll shut up now as I'm off topic.It's been so long I can't remember - seriously. I recall the first build with MAD and seem to remember it had some issues with extraneous clicks at the end of tracks. That was fixed but I can't remember if that was before or after Marcus had debauched the playlist.

Sorry mon...

As to sound quality, QMP easily meets that requirement with MAD. However, I do wish I had no need for such a plugin and seamless / gapless playback of MP3s is very seductive. I wouldn't compromise sound quality for it though which is why I won't use MPG123.

[matty28carter]

No, I'm not going to suggest MAD - licensing issues with the GPL would likely make that quite untenable.

I am going to suggest BASS, which is the heart of several rival players and to my ears every bit as good as MAD. It is a 32-bit decoder and can be paired with dithering and soundshaping a la MAD and is even tweakable for seamless gapless MP3 playback. The author of XMPlay created and licenses it:

www.un4seen.com (http://www.un4seen.com)

It is decidedly better than the outdated and clunky XAudio decoder QCD currently uses from an audio quality perspective and with this engine the player would completely obviate any need for plugins such as MAD and <shudder> MPG123.

Paul?

I've not tried XMPlay (downloading it now) but seeing as their decoder is better (according to roj) and we could use it, why not?

Edit --- Sounds pritty good!

Also, correct me if i'm wrong but - if we do use bass after reading this
BASS is an audio library for use in Windows software. It's purpose is to provide developers with the most powerful and efficient (yet easy to use), sample, stream (MP3, MP2, MP1, OGG, WAV, custom generated, WMA & CD via the add-ons), MOD music (XM, IT, S3M, MOD, MTM, UMX), MO3 music (MP3/OGG compressed MODs), and recording functions. All in a tiny DLL, under 100KB in size.
it'll support MP3, Ogg, MP1/2 and WAV and possibly WMA all in one plug-in thats only 100Kb? Or have i missed something.

[Quake]

I downloaded vuplayer (http://www.vuplayer.com/vuplayer.htm) which uses the Bass Decoder and I have to say the sound is better than QCD's and Winamp. The sound is much clearer and the Bass is much more powerful.

Whatever Paul Chooses, I hope it's gonna be better than Xaudio.

[Paul]

Hey Guys,

I took a look at Bass. Seems it's just a wrapper for DirectShow which means it's using DirectShow filters that are available in Windows. This means it's using the exact same decoders as Windows Media Player. Not that that's a bad thing, it's just not anything new either.

I (or anyone) could make a DirectShow input plugin that does audio. That may be a great idea, seeing as all kinds of crazy codecs are available for DS.

[Roj]

Hey Guys,

I took a look at Bass. Seems it's just a wrapper for DirectShow which means it's using DirectShow filters that are available in Windows. This means it's using the exact same decoders as Windows Media Player. Not that that's a bad thing, it's just not anything new either.

I (or anyone) could make a DirectShow input plugin that does audio. That may be a great idea, seeing as all kinds of crazy codecs are available for DS.
What about the 32-bit decoder for MP3s that's inherent in BASS?

[Paul]

What about the 32-bit decoder for MP3s that's inherent in BASS?
Must be an MP3 DirectShow filter in Windows that supports decoding into 32-bit.

There's two reasons why BASS is just a wrapper:
1) it's only 100kb in size
2) they say so (in a round about way)

[Roj]

Must be an MP3 DirectShow filter in Windows that supports decoding into 32-bit.

There's two reasons why BASS is just a wrapper:
1) it's only 100kb in size
2) they say so (in a round about way)
Hey, as long as it works. :) I haven't checked out Media Player for audio since version 8 so I have no clue how it sounds today. I do know that SP1 for XP overhauled the audio API because M-Audio had a hand in it (read: they redesigned KMIXER for Microsoft). If the wrapper decodes at 32-bit and then uses DirectSound for output with his own customized dithering and sound shaping, that's cool - the hybrid sounds great.

So, you still haven't said - how about it?

[Toe]

Could be interesting. I think I read somewhere that WMP10 comes with an *updated* Fraunhofer decoder. (or was it in SP2?) I know MAD & LAME are both available for dshow. Moonlight Odio DShow can do MP3 & AC-3. Not to mention, a DShow plugin would provide an alternate decoder for things like FLAC and AAC.

By the way, GraphEdit is supposed to be quite handy for figguring out exactly what DShow filters are actually in use. You can get a copy of it from Doom9 (http://www.doom9.org/index.html?/software2.htm).


Now that I think of it, would it be possible to make a DShow encoder plugin? :)

[Roj]

Could be interesting. I think I read somewhere that WMP10 comes with an *updated* Fraunhofer decoder. (or was it in SP2?) I know MAD & LAME are both available for dshow. Moonlight Odio DShow can do MP3 & AC-3. Not to mention, a DShow plugin would provide an alternate decoder for things like FLAC and AAC.

By the way, GraphEdit is supposed to be quite handy for figguring out exactly what DShow filters are actually in use. You can get a copy of it from Doom9 (http://www.doom9.org/index.html?/software2.htm).


Now that I think of it, would it be possible to make a DShow encoder plugin? :)
I personally wouldn't touch FhG *stuff* with *your* 50 foot pole. :) :) :) And yes, it could be very interesting.

[Toe]

I personally wouldn't touch FhG *stuff* with *your* 50 foot pole. :) :) :) And yes, it could be very interesting.
ROL!

Roj, if Paul is right that BASS is just wrapping the standard DShow MP3 filter, then BASS is Fraunhofer. :P

The guys who invented MP3, nah, they know nothing about it. :rolleyes:

[Roj]

ROL!

Roj, if Paul is right that BASS is just wrapping the standard DShow MP3 filter, then BASS is Fraunhofer. :P

The guys who invented MP3, nah, they know nothing about it. :rolleyes:They've proven that well enough with their encoder - LAME flatlined it YEARS ago. Also Winamp has been using FhG for several revs too - and it stank compared to MAD.

Your point? :)

Someone invented something - others improved upon it significantly. Not an unheard of occurrance - just look at Japan and a little thing called "the car".

Also, did FhG decode using 32-bit math? We only started to rise above the norm with MAD as far as I know and it uses 24-bit integer math coupled with a dithering algorithm. This decoder purports to use 32-bit math for decoding calculations - and XMPLay also has dithering and soundshaping built in.

[Toe]

WAV/MP3/MP2/MP1/OGG samples
Supports mono/stereo 8/16 bit samples
From what I can tell, it's like this: it uses the FhG decoder built into Windows and decodes at 16bit. Then *afterward* it resamples the 16bit data to 32bit. Big whoop, a DSP plugin for QCD could do the same thing for QCD's current decoder. Doesn't change the fact that you're trying to get 32bit quality from a source that's far below pure 16bit PCM in quality. (namely, your MP3s)

[Roj]

From what I can tell, it's like this: it uses the FhG decoder built into Windows and decodes at 16bit. Then *afterward* it resamples the 16bit data to 32bit. Big whoop, a DSP plugin for QCD could do the same thing for QCD's current decoder. Doesn't change the fact that you're trying to get 32bit quality from a source that's far below pure 16bit PCM in quality. (namely, your MP3s)You don't understand - I'll try and explain.

It has nothing to do with the sample rate.

It has everything to do with the decoding precision.

If you use 16-bit math precision, you'll get one result. If you use 24-bit precision, you'll get a more accurate result. If you use 32-bit, it will be even more accurate.

More accuracy == more information.

More information == more detail in the sound.

Remember, MP3s are mathematical digital representations of a sound - they're not an analog bitmap of that sound. They get translated to one.

[Toe]

You don't understand.
It has nothing to do with the sample rate.
Did I even mention sample rates anywhere?

It has everything to do with the decoding precision.
And sample rate is somehow unreleated to sound quality?

Remember, MP3 are mathematical digital representations of a sound - they're not an analog bitmap of that sound. They get translated to one.
I'll buy that.

More accuracy == more information.
If we're comparing apples to apples (ie pure PCM to pure PCM) then it's generally necessary.

More information == more detail in the sound.
Yep, if I overlay an extra 16bits of noise over the original, there's gonna be an awful lot more 'detail'. Resampling does not recreate lost detail, it can only take a guess at it. It's slightly different if an MP3 is decoded directly to 32bit, but not much.

If you use 16-bit math precision, you'll get one result. If you use 24-bit precision, you'll get a more accurate result. If you use 32-bit, it will be even more accurate.
Sure. Too bad the BASS 'decoder' you're advocating is only 16bit, eh?

[hedge]

I think Toe's got a good point. If the source has already been decoded into 16bit, then that sample is being upsampled to 32bit, where is the gain gonna come from? The 16bit math precision results you speak of would have already been produced.

Sounds all to much like the transcoding scenario to me.

But does bring up another good point, hell I don't have a 24bit soundcard, and whilst i might eventually get one, I don't think I would be able to make out any difference between that and 16bit.

[Todd The Kiwi]

Someone invented something - others improved upon it significantly. Not an unheard of occurrance - just look at Japan and a little thing called "the car".

wankel engine
portable cassette player
sushi,no wait,that's theirs... :cheeky:

[Tokelil]

I would have thought all decoders uses 32 bit internal precision. MAD uses 32 bit interger.

[MLO]

Hey Guys,

I took a look at Bass. Seems it's just a wrapper for DirectShow which means it's using DirectShow filters that are available in Windows. This means it's using the exact same decoders as Windows Media Player. Not that that's a bad thing, it's just not anything new either.

I (or anyone) could make a DirectShow input plugin that does audio. That may be a great idea, seeing as all kinds of crazy codecs are available for DS.
I was very surprised to see talk of Ian Luck's BASS project here, it has been a favorite among Delphi programmers for years. And it is a little more than just a wrapper, but calling it a wrapper with respect to most of it's main features is definely NOT out of line. Ian gets the 100K file from this little program he wrote: http://un4seen.com/petite/

Anyway this is all beside the point, it would not be very difficult to wrap the bass.dll in a QCD input plugin... I screwed around with the Winamp input plugin for BASS (it only handles MOD formats) and added MP3 playback to it, and then used the WPM (winamp plugin manager) to get playback in QCD. Not the most elegant solution, but it worked (although I didn't hear any difference in audio quality - tell you the truth I can't tell the difference among any of the MP3 decoders out there today).

My only concern with QCD's current MP3 plugin is that it can't handle a lot of the "Shoutcast" streams out there today. Why? Because they aren't MP3 streams... some are AAC LC and some are AAC plus (via AOL's ultravox servers - uvox://). For this reason alone, I have been using Winamp's in_mp3.dll for my playback in QCD. After all, who cares about sound quality if you can't even hear it. :)

[Rex_Mundi_Incarnit]

About replacing the default mp3 encoder, i dont think the encoder is that bad. Besides there is also the lameMp3 encoder for QCD and i thought that lameMp3 was top of the world. If i can encode mp3s at a sample rate of 44.1kHz, at a bit rate of 320 kbps, and at 16bit, i dont know how sound can get any clearer. It all depends on the source (like cd's which sample at 44.1kHz anyways), speakers (most people have standard desktop speakers and claim they hear a difference in sound between qcd and winamp, yeah right [sarc]), and soundcard/chip. Of course, its always fun and attractive to get the latest mp3 encoding technology into a media player, but sometimes the need for it is like upgrading from wmp 8 to wmp 9. I.e. cool, done that, now what am i supposed to notice that is different in sound quality? :confused:

[hedge]

About replacing the default mp3 encoder, i dont think the encoder is that bad. Besides there is also the lameMp3 encoder for QCD and i thought that lameMp3 was top of the world. If i can encode mp3s at a sample rate of 44.1kHz, at a bit rate of 320 kbps, and at 16bit, i dont know how sound can get any clearer. It all depends on the source (like cd's which sample at 44.1kHz anyways), speakers (most people have standard desktop speakers and claim they hear a difference in sound between qcd and winamp, yeah right [sarc]), and soundcard/chip. Of course, its always fun and attractive to get the latest mp3 encoding technology into a media player, but sometimes the need for it is like upgrading from wmp 8 to wmp 9. I.e. cool, done that, now what am i supposed to notice that is different in sound quality? :confused:
ummm, I think you got a little confused... or else i am ;) but weren't we talking about decoders, not encoders?

[Roj]

Did I even mention sample rates anywhere?


And sample rate is somehow unreleated to sound quality?


I'll buy that.


If we're comparing apples to apples (ie pure PCM to pure PCM) then it's generally necessary.


Yep, if I overlay an extra 16bits of noise over the original, there's gonna be an awful lot more 'detail'. Resampling does not recreate lost detail, it can only take a guess at it. It's slightly different if an MP3 is decoded directly to 32bit, but not much.


Sure. Too bad the BASS 'decoder' you're advocating is only 16bit, eh?Dead wrong on almost all counts.

Do not compare a bitmapped PCM file to an MP3 - your first mistake. One is a 16-bit PCM wav analog IMAGE and the other is a 32-bit math digital representation of that image - not the same thing at all. Go and read what an MP3 actually is before commenting - your entire train of thought is off base, albeit a very common mistake.

Better yet, go to Rob Leslie's site and read how MAD works.

[Roj]

I think Toe's got a good point. If the source has already been decoded into 16bit, then that sample is being upsampled to 32bit, where is the gain gonna come from? The 16bit math precision results you speak of would have already been produced.

Sounds all to much like the transcoding scenario to me.

But does bring up another good point, hell I don't have a 24bit soundcard, and whilst i might eventually get one, I don't think I would be able to make out any difference between that and 16bit.See my take on Toe's "logic" - he's way off base. A 24-bit sound card will ALWAYS sound better than a 16-bit sound card, all other things being equal (quality algorithms, good DACs, a straight 24-bit data path bereft of resamplung) primarily because the noise floor will be lower.

As an aside, Creative Labs cards fail on all three points compared to the competition.

[matty28carter]

On un4seen.com it says for the bass engine Sample streams
Stream any sample data in 8/16/32 bit
this is the 32 bit we're all on about. Right?

Ug. Now i'm confused.

WAV/MP3/MP2/MP1/OGG samples
Supports mono/stereo 8/16 bit samples (including compressed WAVs)

[Roj]

No - those refer to bitmapped streams.

We're talking about decoder precision (at least I am - Toe doesn't seem to get it).

Calculate pi to 10 decimal places.

Now calculate pi to 20 decimal places.

Which result is more accurate?

That pretty much covers it.

[Toe]

Roj, are you really that slow? My point is that BASS, which you came in here recommending in the first place, is NOT a 32bit decoder at all! It's just a wrapper around the humble 16bit Fraunhofer decoder (which you later trashed) that BASS can then resample to 32bit after it has been decoded to 16bit. You're not going straight from MP3 -> 32bit PCM, you're going from MP3 -> 16bit PCM -> resampled 32bit. Big difference! If that's what you want, we may as well just have a DSP plugin that resamples the output of the curent decoder and forget this "replace the default MP3 decoder" business, as the result would be about the same.

Course, in all this talk, we've left actual audible gains in sound quality out of this. But of course, I know you'd never actually put your money where your mouth is and take a test, Roj. ;)

Until then, it's all placebo.

[fatal error]

< develops a prominent unibrow and tries to re-invent the wheel.

[brian]

This discussion has become too technical. The basic questions are whether an improvement on the present QCD default mp3 plugin is (1) available and (2) of interest to Paul. His intervention in this thread suggests that neither (1) nor (2) is the case.

[Roj]

Roj, are you really that slow? My point is that BASS, which you came in here recommending in the first place, is NOT a 32bit decoder at all! It's just a wrapper around the humble 16bit Fraunhofer decoder (which you later trashed) that BASS can then resample to 32bit after it has been decoded to 16bit. You're not going straight from MP3 -> 32bit PCM, you're going from MP3 -> 16bit PCM -> resampled 32bit. Big difference! If that's what you want, we may as well just have a DSP plugin that resamples the output of the curent decoder and forget this "replace the default MP3 decoder" business, as the result would be about the same.

Course, in all this talk, we've left actual audible gains in sound quality out of this. But of course, I know you'd never actually put your money where your mouth is and take a test, Roj. ;)

Until then, it's all placebo.You still persist in confusing bitmapped wav files with calculation precision.

Tell you what:

take it up with the author and quit wasting my (and everyone's) time with erroneous "information".

Do note that if FhG *were* being used, the decoder would likely NOT be 32-bit precision. All my experience wit FhG indicates that they never used that depth of mathematical calculation - they likely never felt it was worth it and after all it was developed way back when at a time when high resolution consumer sound cards which could benefit from that precision (let alone dithering and soundshaping) were not even conceived of. It would be interesting to see you wriggle out of that one but I hardly think it's worth the entertainment value.

An additional tip:

You'd do well to stop parroting Hydrogen Audio - it's old and as time has shown, there's often more "religion" there than beef.

On that note, I think I'm done withthis particular discussion with you - after all, you can bring a horse to water but you can't make them drink and recognizing when further debate is pointless is one of the tenets of quality debating.

[Toe]

You still persist in confusing bitmapped wav files with calculation precision.
Not at all. 32bit internal precision can be a very good thing, if 1) if the decoder actually has 32bit internal precision in the first place, 2) the decoder is actually doing something with those extra bits, such as dithering. As the MAD website states,

Even when the output device supports only 16-bit PCM, applications can use the extra resolution to increase the audible dynamic range through the use of dithering or noise shaping.

But that dithering and noise shaping needs to be done before it gets converted back into 16bit PCM. Otherwise, all that extra calculation precision on the MP3's 'representation' of the audio is lost.

Sure, you can have 32bit internal calculation precision for things later in the chain (DSPs), but that doesn't make an MP3 decoder with 16bit internal precision into an MP3 decoder with 32bit internal precision.


By the way, you still haven't addressed the fact that you dissed the Fraunhofer decoder, yet you're still cheering BASS which is just a wrapper for the Fraunhofer decoder. ;)

[Toe]

Whoa, nice edit of your post after I've already posted my reply to it there, buddy! :)

[Tokelil]

Roj: For us "uneducated", what do you mean by a bitmapped wave?

[Roj]

Whoa, nice edit of your post after I've already posted my reply to it there, buddy! :)
I hadn't bothered to read your reply at the time - it wasn't worth it. I did add things I considered relevant that I hadn't put in the first time (I'm in the process of rebuilding my primary box with a massive hard drive upgrade - 120gb and 250gb - so you'll have to excuse me if I'm not on my typically precise form).

As to not addressing the issue you stated, I most certainly have - I put the onus back on *you* to take it up with the BASS *author*. After all, HE is the definitive source on his own decoder I should think.

[Rex_Mundi_Incarnit]

ummm, I think you got a little confused... or else i am ;) but weren't we talking about decoders, not encoders?

I kinda missed the point, yes. Decoders, not encoders. I still stick to the point tho that the default mp3 decoder is great enough for me and that for the average windows user it should be too. It has stream saving, shoutcast titles, does what it's supposed to do w/h crashing and sounds better than winamp (to some, or im really getting confused here). I shall not interfere in a technical discussion again, cus this is what happens, i got confused :) As long as i get good sound from my speakers im happy ;)

[Tokelil]

The author of BASS only claims 16/24/32 bit output precision as far as I remember...

[Qaz]

I hadn't bothered to read your reply at the time - it wasn't worth it.
No a fence but this gotta be the worst reply ever.

[Roj]

Roj: For us "uneducated", what do you mean by a bitmapped wave?
Think of a wav file as essentially a bitmapped analog image of a sound.

A MP3 takes that bitmap and does a lossy reduction on it to produce a file which represents much, but not all, of the information contained in the original.

That file is not an analog image. It is a sequence if you will that has to be run through a series of calculations to decode the file into another analog bitmap - but NOT the original analog bitmap. Portions are missing because they have been removed by the encoding process. The decoding can take place at varying levels of calculation precision. Those levels of calcualtion precision have zero to do with the original way the wav was recorded (8-bit / 16-bit / 24-bit).

For a better explanation, check out Rob Leslie's site - he's the author of MAD.

[Tokelil]

I know the teori behind wave files, lossy intepretations of wave files etc. Just never heard it called a bitmapped wave... I guess what you mean by that is just that it is a digital representation of a anolog wave.

[Roj]

I know the teori behind wave files, lossy intepretations of wave files etc. Just never heard it called a bitmapped wave... I guess what you mean by that is just that it is a digital representation of a anolog wave.
A MP3 is a digital representation of an analog wav with pieces missing which must be calculated back into an analog representation (decoded) before it can be dealt with my an analog playback system. A wav file just gets dumped to the analog playback system - no extra step needed.

[Toe]

OK, it's been 2 hours now. Can I reply to what you've written or are you still changing your story?

As to not addressing the issue you stated, I most certainly have - I put the onus back on *you* to take it up with the BASS *author*. After all, HE is the definitive source on his own decoder I should think.
Roj, YOU started a thread saying that QCD's default MP3 decoder needs to be replaced with BASS. If that's your goal, then YOU have the burden of proof here. YOU have to convince me, the other users of QCD, and most importantly Paul that BASS is both truly better than the current decoder, and worth the development time to switch everything over. If that's what you're trying to do with this thread, then trying to pass that job over to me will get you precisely nowhere.

Your reply still amounts to another dodge. Are you hoping that BASS's author will be able to explain why Fraunhofer's decoder sounds like trash to you, yet a wrapper around Fraunhofer's decoder somehow sounds wonderful to you?

I'll give you a hint:
PLAAA-CEEEE-BOOO ;)

[Roj]

Roj, YOU started a thread saying that QCD's default MP3 decoder needs to be replaced with BASS. If that's your goal, then YOU have the burden of proof here.No lad. I had all the proof I needed:

It Sounded Better To My Ears (and those of others too when it comes to that).

That's all the proof I need and 'nuff said.

I'll give you a hint:
PLAAA-CEEEE-BOOO ;)I'll give you a more valid one:

Empty Hackneyed Rhetoric With No Backup.

By the way, have you talked with Ian yet to clarify things?

No?

Why am I not even the tiniest bit surprized.

You've had a marked tendency over time to spout buzzwords (open source, placebo and of course your favorite: ABX), blow smoke and do little else.

It won't surprize me if you never deliver on this. But hey, if you do and I'm wrong I'll admit it - at least I'll actually have the good sense to be able to tell when I'm off-base. :) :) :)

Here's how it works:

If I see a reply from you, I quickly scan it. That generally doesn't take a great deal of time given the usual content. If there is no meaningful information contained therein (read: you're invoking your standard rhetoric), I'll pick a key phrase and comment on it, usually (re)iterating a request for backup supporting whatever it is you're emoting about.

The rest gets the bit bucket.

That generally has a tendency to save a lot of my time.

On that note:

g'nite.

Let me know what Ian says in the morning.

[Roj]

From what I can tell, it's like this: it uses the FhG decoder built into Windows and decodes at 16bit. Then *afterward* it resamples the 16bit data to 32bit. Big whoop, a DSP plugin for QCD could do the same thing for QCD's current decoder. Doesn't change the fact that you're trying to get 32bit quality from a source that's far below pure 16bit PCM in quality. (namely, your MP3s)Toe:

You're still not getting it.

The PCM data is indeed 16-bit - I'm not disputing that.

The mathematical precision converting it from mp3 (which is NOT native PCM) is NOT 16-bit.

You keep confusing 16-bit PCM with 16-bit math precision. The only thing those two items have in common is the number 16!!!!!

I finally found an e-mail message from Rob Leslie (author of MAD) dated April 12th, 2003 that illustrates this:

____________________

> I have a 16-bit Turtle Beach Santa Cruz. If I select 24-bit decoding
> in the Winamp plugin, what exactly occurs? Is it a case of the
> decoder calculating to 24-bit precision and then dithering down to
> 16-bit? Or is that setting directly related to the sampling rate of
> the sound card?

Hi Roj,

The resolution you select in the plug-in is the resolution sent to the
sound driver for your hardware. What happens there depends on your
driver and, sometimes, your operating system.

Most likely what happens when you select 24-bit resolution for hardware
that only supports 16 bits is the extra resolution is truncated or
rounded. Therefore you will probably get better quality output by
selecting 16-bit resolution from the plug-in, as then the audio will be
dithered from the internally decoded high resolution to 16 bits by the
plug-in before sending to the sound driver.

____________________

The key words in that answer are "internally decoded high high resolution".

NOW do you understand?????????? (mops brow)

Now I really gotta go - there's a pitcher of Pina Coladas made with Dominican Gasoline (150 proof Brugal rum) with my name on it waiting for me.

Have a good nite...

[Toe]

You're still lashing out at straw men here, Roj. I fully understand the difference between internal calculation precision and output bit depth. I was only trying to state that there's a difference between either of those two and a program that takes 16bit input (from an MP3 decoder or whatever else it may be), resamples that to 32bits and passes it to DSP plugins, and calls it "32bit signal processing". (You always have to take what the marketing department tells you with a grain of salt, unless you're ready to believe MS's claim that 64k WMA = 128k MP3 ;)) What exactly is it that you get with BASS? I don't know and frankly right now I don't care.


-----------------
Look, Roj, this isn't going to get either of us anywhere. What do you say we at least try to argue something a little more productive here?
-----------------


I think there's a few questions that need to be asked before we jump straight to "Let's replace QCD's MP3 decoder with BASS!"

1) Is there even anything wrong with QCD's existing MP3 decoder?
Given the first few replies to this thread, I think it at least merits consideration that there is.

2) What exactly are the problems, if there are any?
I think there's one that we can all agree on: gapless playback. QCD's built-in handling of that isn't quite seamless. Let's try to leave subjective sound quality out of this, down that road lies madness, at least as far as this thread goes. Let's also leave out the buzzwords like 32bit precision and noise shaping and dithering and other things you may or may not actually be able to be able to HEAR a difference in (http://www.un4seen.com/forum/?board=2;action=display;threadid=3175;start=msg209 70#msg20970). For now at least, let's relegate that stuff to 'fringe benefit' status. Nice to have if it's there, but don't base the whole decision on it.

3) (a biggie that hasn't been touched yet) Can QCD's existing decoder be modified to correct any problems?
Only Paul can answer this one, really.

4) Would it be worth the trouble of replacing the default decoder, and would there be any side effects from removing the current default?
IIRC Paul has stated that equalization of MP3s is performed inside the decoder itself for reasons of performance. You and I may not care about a small performance hit, but we can't speak for everyone who uses QCD here. As for how much trouble it would be in terms of development, again only Paul can answer.

5) Would this be better off as a seperate plugin, rather than as the default?
It's probably easier just to make it available as a seperate download, rather than integrate it as the default. That way, those who want it can grab it, and those who don't can be happy with the default.

However, first impressions can be a bitch. Someone trying QCD for the first time might start it up and check for gapless playback, or whatever feature they look for in an MP3 player, and find that QCD doesn't do it right. They could ditch QCD without even thinking to look for a seperate decoder plugin.

As a corollary here, would a DSP that takes existing output and resamples/interpolates it to a higher bitrate/samplerate be beneficial? As I said earlier, this isn't the same as having a decoder with 32bit internal precision and 32bit output, but some may still find the result pleasing.

6) If it is truly desirable to replace the default decoder, what are the contenders?
MAD was initially dismissed due to it being GPL'd, but according to their front page it's also available under a commercial license. Whether or not the terms of that license (or the GPL, for that matter) is acceptable to Paul, again only he can decide.

You wrote off mpg123, but apparently you like LAME, which as they say right on their front page (http://lame.sourceforge.net/) uses mpg123 as its decoder. Apparently it's LGPL'd (http://www.mpg123.de/), which AFAIK means that it can be linked with non-open source binaries. (ie, would be friendlier than the GPL if Paul wishes to keep QCD closed-source)

By the way, don't forget that there's a fuckload lot of variations on mpg123. Supposedly Otachan's (http://www3.cypress.ne.jp/otachan/) tweaked out version of Shibatch's mpg123 for WinAMP has excellent gapless support. :) EDIT: Come to think of it, wasn't this the one used in the 'eye of newt' method?

There's the Moonlight Odio Dekoda (http://www.elecard.net.ru/download/) (dig that name) I mentioned earlier, but I can't find any sort of license info on it. They make the DShow filter available as a free download though, so if we got the DShow plugin for QCD we could use it that way. I also note that it apparently has built-in AC-3 handling, a nice one for the 'fringe benefits' category.

Of course, there's your BASS suggestion.

We could always go the Fraunhofer route. :P

There's probably several others out there that I don't know about...

7) Are there any fringe benefits with any alternate decoder?
Possiblities include increased performance, smaller footprint, better portability (ie if Paul ever changes his mind and makes a version of QCD for Linux or Mac, is there a version of this lib already available?), etc. Would take quite a bit of research into each option.


Any other thoughts?

[brian]

Does the notion of a default mp3 decoder have any real meaning here, when any QCD plugin can be enabled or disabled at will? I always thought the whole basis of QCD was that users can choose for themselves which of the available decoders to use. So the issue is that of whether to widen the choice, which has to be a good plan - provided someone is willing and able to do the development work.

[Roj]

Quick scan, no beef from Ian. I almost moved on and said "next" :) ...

...but you raised some reasonable questions. Here are my takes:


"1) Is there even anything wrong with QCD's existing MP3 decoder?"

Yes. Compared to MAD and BASS it's a poor third, tying with FhG and (shudder) mpg123 which is effectively a FhG clone and nothing more. All of those decoders have a similar sound characterized by a lack of depth, harshness and tinniness with subtle details getting lost in the shuffle. In fact, I refused to consider QCD until MAD became available for it.

"2) What exactly are the problems, if there are any?"

See above and add occasional miscalculation of track times, no dithering, no soundshaping and no seamless / gapless mp3 playback.

"3) (a biggie that hasn't been touched yet) Can QCD's existing decoder be modified to correct any problems?"

No. Its a package from XAudio - you pays your money and you takes your chance. CoolPlayer once used XAudio. It was replaced with MAD.

"4) Would it be worth the trouble of replacing the default decoder, and would there be any side effects from removing the current default?"

That only Paul can answer. From an audio quality perspective, I vote a resounding "yes". However, I don't have to do the coding now, do I... :)

"5 Would this be better off as a seperate plugin, rather than as the default?"

Given our current scenario with plugins? Are you serious?

"6) If it is truly desirable to replace the default decoder, what are the contenders?"

Good question. If MAD were purchased (if I were doing it, the price would have to be equal to or more reasonable than that being charged for BASS) then that would be cool. BASS is another contender. As I pointed out above, I see no other valid contenders out there from an audio quality perspective. High resolution decoding, dithering, soundshaping and seamless / gapless playback are where it's at. Winamp, XMPlay, CoolPlayer and even fubar all have those features in one form or another although some have it via plugins.

So there it sits...

[brian]

Anyway this is all beside the point, it would not be very difficult to wrap the bass.dll in a QCD input plugin... I screwed around with the Winamp input plugin for BASS (it only handles MOD formats) and added MP3 playback to it, and then used the WPM (winamp plugin manager) to get playback in QCD. Not the most elegant solution, but it worked (although I didn't hear any difference in audio quality - tell you the truth I can't tell the difference among any of the MP3 decoders out there today).
hj - A couple of requests for you, please:

(1) In what other ways, if any, does the Winamp in_mp3.dll differ technically from the QCD counterpart (i.e. leaving aside audible sound quality)?

(2) Could you possibly make your mp3-enabled Winamp BASS plugin available for download?

Many thanks

[Roj]

Any other thoughts?Additional info from Ian (creater of BASS and XMPlay), taken directly from a post by him on the XMPlay forum:



XMPlay doesn't use any DirectX stuff, but it's still a very long way away (too far http://www.un4seen.com/forum/YI/wink.gif) from being a simple "recompile under Linux" job. It'd be a lot simpler if WineLib could do most of the porting job. Apparently, WineLib has been used to port some large commercial apps, but as I say, I can't get it to do anything.The context of this was someone asking him to port XMPlay to Linux. If I read this correctly, this puts paid to the whole concept that BASS is a simple wrapper for native FhG functionality built into Windows.

[deusdiabolus]

You have argued this point so fervently, I think the ultimate solution would be for you to code a decoder plugin for QCD using the BASS engine. Perhaps then, if everyone tries it in QCD and really likes it, Paul would replace the default plugin with it.

Outside of that, I think the request has been adequately stated and argued.

[Roj]

You have argued this point so fervently, I think the ultimate solution would be for you to code a decoder plugin for QCD using the BASS engine. Perhaps then, if everyone tries it in QCD and really likes it, Paul would replace the default plugin with it.

Outside of that, I think the request has been adequately stated and argued.I'm not and have never claimed to be a coder.

Audio is one of my passions, not codespinning.

However, I do reserve the right to indicate what sounds good to me and what doesn't and have never been shy about doing so. :)

One day I'll have to relate the story of what happened the first time I heard the Klipsch Pro Medias - let's just say the salesperson wasn't expecting quite what he got. :) :) :)

[MLO]

hj - A couple of requests for you, please:

(1) In what other ways, if any, does the Winamp in_mp3.dll differ technically from the QCD counterpart (i.e. leaving aside audible sound quality)?

(2) Could you possibly make your mp3-enabled Winamp BASS plugin available for download?

Many thanks
Sorry Brian, I stopped paying attention to this thread after that long debate started. http://www.quinnware.com/forum/images/emots/post/cyclops.gif

1) As I stated in my earlier post, Winamp's in_mp3 plugin also handles AAC files and AAC (LC and plus) streams. These AAC encoded streams can seamlessly appear as Shoutcast streams since the same plugin handles them. In addition the in_mp3 plugin gets used by in_mp4 for decoding AAC files wrapped in the MP4 container. It also gets used by in_dshow and in_nsv for playing back the audio portion (when it is AAC or MP3) of various formats.

2) I really doubt you want it... keep in mind that the plugin didn't even process the audio through Winamp, therefore the only thing it does is play... nothing else works as it should (no vis/dsp/output plugin functionality).

[brian]

1) As I stated in my earlier post, Winamp's in_mp3 plugin also handles AAC files and AAC (LC and plus) streams. These AAC encoded streams can seamlessly appear as Shoutcast streams since the same plugin handles them. In addition the in_mp3 plugin gets used by in_mp4 for decoding AAC files wrapped in the MP4 container. It also gets used by in_dshow and in_nsv for playing back the audio portion (when it is AAC or MP3) of various formats.

Thanks for this. What I was hoping to find out, please, is whether the mp3 decoder in the Winamp plugin is based on the same decoding technology as the one in the QCD plugin. Apologies if I didn't tmake this clear.

[Roj]

Thanks for this. What I was hoping to find out, please, is whether the mp3 decoder in the Winamp plugin is based on the same decoding technology as the one in the QCD plugin. Apologies if I didn't tmake this clear.
Winamp uses FhG and has for several releases now. QCD uses XAudio. Perhaps that was what you were driving at?

[brian]

Winamp uses FhG and has for several releases now. QCD uses XAudio. Perhaps that was what you were driving at?
Yes, thanks. And what's the difference between them, broadly speaking, please?

[Toe]

Does the notion of a default mp3 decoder have any real meaning here, when any QCD plugin can be enabled or disabled at will? I always thought the whole basis of QCD was that users can choose for themselves which of the available decoders to use. So the issue is that of whether to widen the choice, which has to be a good plan - provided someone is willing and able to do the development work.
See #5 in my post above. Basicly, consider the first-time QCD user who may not bother trying plugins before passing judgment. You'd want to make as good an impression as possible with the program's core functionality (ie how well it plays MP3s).

One day I'll have to relate the story of what happened the first time I heard the Klipsch Pro Medias - let's just say the salesperson wasn't expecting quite what he got.
Heh. Nice and punchy, but a wee bit harsh, IMHO.

Winamp uses FhG and has for several releases now.
Just for a bit of extra clarification, once upon a time WinAMP also had its own (buggy) MP3 decoder by the name of Nitrane. IIRC that was abandoned with the 2.7 release. Originally, WA used the AMP decoder, hence the name.

Yes, thanks. And what's the difference between them, broadly speaking, please?
When it comes down to it, not a whole lot. MP3 decoding is very rigidly defined in the standard. (Unlike MP3 encoding, which can be done a million ways.) If two decoders decode an MP3 differently, one (or both) of the decoders is buggy. One potential difference is in how they handle gapless MP3 playback. You can at least somewhat deal with this problem with a gap-removing output plugin (as QCD currently does), but this is often not seamless, leading to a small click or pop between tracks. Another possible difference with a few decoders is extra processing to the output, such as dithering. This might make a difference in the sound quality, but... well, let's just say you're probably better off not paying attention to the buzzwords (http://www.un4seen.com/forum/?board=2;action=display;threadid=3175;start=msg209 70#msg20970), shall we? :)

[Toe]

By the way, is it verified that BASS itself includes gapless MP3 playback? I notice that gapless playback is placed very prominently on XMPlay's feature list, yet it isn't mentioned on the features page for BASS at all...

[brian]

When it comes down to it, not a whole lot. MP3 decoding is very rigidly defined in the standard. (Unlike MP3 encoding, which can be done a million ways.) If two decoders decode an MP3 differently, one (or both) of the decoders is buggy. One potential difference is in how they handle gapless MP3 playback. You can at least somewhat deal with this problem with a gap-removing output plugin (as QCD currently does), but this is often not seamless, leading to a small click or pop between tracks. Another possible difference with a few decoders is extra processing to the output, such as dithering.
Toe - thanks, that's very clear. As you may have gathered, I've only just got to the stage of being aware of the existence of the various decoders, and of which player uses which decoder.

The Xing decoder, as used formerly by FreeAmp and now by Zinf, produces results which sound pretty good to my ears. It's on GPL, too. I have no idea how it differs technically from the other decoders. Any thoughts?

http://www.zinf.org/about.php

[Roj]

Toe - thanks, that's very clear. As you may have gathered, I've only just got to the stage of being aware of the existence of the various decoders, and of which player uses which decoder.

The Xing decoder, as used formerly by FreeAmp and now by Zinf, produces results which sound pretty good to my ears. It's on GPL, too. I have no idea how it differs technically from the other decoders. Any thoughts?

http://www.zinf.org/about.phpLast I checked, Xing was the LEAST accurate of all the decoders out there (there was a table on the web of how accurate the decoding calculations were and Xing did not fare well at all). Their encoder was even worse. The folks at Zinf have long talked about switching to MAD (there's actually an item in their "to-do" list for that) but have never had the resources / cycles to do it.

[Roj]

By the way, is it verified that BASS itself includes gapless MP3 playback? I notice that gapless playback is placed very prominently on XMPlay's feature list, yet it isn't mentioned on the features page for BASS at all...Methinks it's done the same way gapless playback for mp3 is done in any other player, namely by tweaking the buffers. As you likely know, MP3 inherently wasn't designed for gapless but OGG, FLAC and APE were. Even with the latter three formats, I believe work has to be done by the programmer over and above implementing the decoder to make this happen.

[Roj]

Heh. Nice and punchy, but a wee bit harsh, IMHO.I positively hated them. All highs and boom (the only Klipsch multimedia sub I liked was the one that came with the 5.1 system - it didn't rattle at high volumes and was actually tight) and no mids. They had zero soundstage or imaging - totally in-yer-face FLAT. The salesperson expected gushing praise (after all, most kids hearing them practically peed themselves with joy) and what he got was "these are hideously overpriced for what you get - they're great for games and movies but absolute crap for music". I then proceeded to point out exactly what was missing in the sound. Unfortunately for him he had gathered a few people around for the demo. This pretty much covers my views on the subject:

http://www.epinions.com/content_6498651780

The 4.1s fixed the midrange issue but those half-assed mini horns and overblown subs just annoy my ears.

Just for a bit of extra clarification, once upon a time WinAMP also had its own (buggy) MP3 decoder by the name of Nitrane. IIRC that was abandoned with the 2.7 release. Originally, WA used the AMP decoder, hence the name.I remember it well. Pure unadulterated garbage.

When it comes down to it, not a whole lot. MP3 decoding is very rigidly defined in the standard. (Unlike MP3 encoding, which can be done a million ways.) If two decoders decode an MP3 differently, one (or both) of the decoders is buggy. One potential difference is in how they handle gapless MP3 playback. You can at least somewhat deal with this problem with a gap-removing output plugin (as QCD currently does), but this is often not seamless, leading to a small click or pop between tracks. Another possible difference with a few decoders is extra processing to the output, such as dithering. This might make a difference in the sound quality, but... well, let's just say you're probably better off not paying attention to the buzzwords (http://www.un4seen.com/forum/?board=2;action=display;threadid=3175;start=msg209 70#msg20970), shall we? :)Hey, I don't disagree with you at all on this and I do apologise for lashing out the way I did. The dithering built into MAD, BASS and fubar's decoders do warp the sound somewhat - but so do tube amplifiers in real audio (that lovely "warm" sound is essentially distortion). Dolby B / C / HX / etc. also warped sound. The magnificent dithering algorithms used by the Santa Cruz sound card (a favorite of mine) also warp sound. However, the dithering and sound shaping of various mp3 decoders (or rather the decode post-processing of various solutions) have a tendency to bring out more detail. That's what I'm in love with. It should come as no surprise to you that in my formative years in audio, I was heavily exposed to tube amplifiers and flat panel speakers, specifically Audio Research and Magneplanar kit. They left a lasting impression.

[hedge]

Methinks it's done the same way gapless playback for mp3 is done in any other player, namely by tweaking the buffers. MP3 inherently wasn't designed for gapless but OGG, FLAC and APE were.I thought the newer versions of LAME included some form of native gapless playback?

[Roj]

I thought the newer versions of LAME included some form of native gapless playback?
They do but you still have to play with the output as I understand it.

[Roj]

By the way, is it verified that BASS itself includes gapless MP3 playback? I notice that gapless playback is placed very prominently on XMPlay's feature list, yet it isn't mentioned on the features page for BASS at all...More info on BASS and gapless:

According to Ian via a post in the XMPlay forum, the decoder checks for the header included by LAME 3.90 and higher vis a vis gapless mp3 encoding. If that header is there, appropriate action can be taken.

[Toe]

there was a table on the web of how accurate the decoding calculations were and Xing did not fare well at all
Do you mean this decoder test (http://mp3decoders.mp3-tech.org/contents.html)? Do keep in mind this test dates all the way back to 2001! IIRC the author has stated that he keeps it up mainly because so many pages still link to it, as it's rather outdated at this point. 3 years is quite a while to apply bugfixes & tweaks.

I thought the newer versions of LAME included some form of native gapless playback?
LAME since like v3.80 (don't remember exactly when) have contained a tag in the header that specifies EXACTLY where a track needs to be chopped to provide gapless playback. (As opposed to the guessing game that's normally done by the player.)

[Toe]

You know, I can't believe I didn't think of this before.

As another contender for default MP3 decoder, the source for foobar's souped-up decoder is available with fb2k's SDK (http://www.foobar2000.org/SDK.zip). (Look under foobar2000\foo_input_std\mpglib inside the zip.) It's even available under a closed source linkable (and thus, QCD friendly) LGPL license. :)

[Roj]

Do you mean this decoder test (http://mp3decoders.mp3-tech.org/contents.html)? Do keep in mind this test dates all the way back to 2001! IIRC the author has stated that he keeps it up mainly because so many pages still link to it, as it's rather outdated at this point. 3 years is quite a while to apply bugfixes & tweaks.
I agree - except Xing hasn't. They went the way of the dinosaur soon after.

LAME since like v3.80 (don't remember exactly when) have contained a tag in the header that specifies EXACTLY where a track needs to be chopped to provide gapless playback. (As opposed to the guessing game that's normally done by the player.)
Was it 3.80? I thought it was more recent than that. That aside, the feature you're referring to is exactly what I'm referencing.

[Roj]

You know, I can't believe I didn't think of this before.

As another contender for default MP3 decoder, the source for foobar's souped-up decoder is available with fb2k's SDK (http://www.foobar2000.org/SDK.zip). (Look under foobar2000\foo_input_std\mpglib inside the zip.) It's even available under a closed source linkable (and thus, QCD friendly) LGPL license. :)I'm sorry but I can't bring myself to trust *anything* PP does. After all, he was the one responsible for Winamp's audio subsystem for many a year and from where I'm sitting that ain't nothing to crow about. Then there's the whole concept of implementing a hack / kludge (kernel streaming) as a major audio alternative system but I'd rather not go there...

MAD too was based on that ancient library (mpglib). I'd feel happier about that product than something cooked up because someone felt he had to reinvent the wheel.

[Toe]

Was it 3.80? I thought it was more recent than that.
As I said, I'm not at all certain on exactly what version introduced it, but it was a couple years ago.

I'm sorry but I can't bring myself to trust *anything* PP does. After all, he was the one responsible for Winamp's audio subsystem for many a year and from where I'm sitting that ain't nothing to crow about.
Actually, AFAIK the whole reason he left WinAMP was because they wouldn't LET him all the spiffy changes he wanted to with WA. Result: first WinAMP 3, and then back to the same ol' WinAMP 2 from ages ago. :)

Then there's the whole concept of implementing a hack / kludge (kernel streaming) as a major audio alternative system but I'd rather not go there...
From foobar's FAQ (http://www.foobar2000.org/FAQ.html#FAQ_o2): "Please note that Kernel Streaming is an experimental feature, it has known issues and using it is *not* recommended unless you have a major reason to." It's pretty much a 'there if you REALLY want it' sort of thing, not something they seriously suggest you use "as a major audio alternative system."

I'd feel happier about that product than something cooked up because someone felt he had to reinvent the wheel.
How can you differentiate between MAD 'reinventing the wheel' and foobar doing the same? Frankly, neither one of us has the technical knowledge to say that one is truely better than the other. Assuming the implementation is good (proper streaming support & such) I'd be quite happy to have either one of them as QCD's default MP3 decoder. :)

[Roj]

From foobar's FAQ (http://www.foobar2000.org/FAQ.html#FAQ_o2): "Please note that Kernel Streaming is an experimental feature, it has known issues and using it is *not* recommended unless you have a major reason to." It's pretty much a 'there if you REALLY want it' sort of thing, not something they seriously suggest you use "as a major audio alternative system."That's there NOW. Initially it most certainly was NOT and was trumpeted as the second coming. I was on 3DSS at the time and decried it vehemently for the sham it was (I had professional contacts within the kernel design team at MS). I also proceeded to educate as many folks as I could as to the mess and quoted Microsoft's own knowledgebase on the subject where they freely admitted that a) it was a kludge b) it wasn't supported and c) it was to be phased out with Longhorn. Finally the fubarians bowed to the pressure of overwhelming evidence with the result that you posted.

Suddenly the emperor had no new clothes.

How can you differentiate between MAD 'reinventing the wheel' and foobar doing the same?Simple.

Rob Leslie created MAD years before fubar was even a thought in PP's mind - in fact it debuted while LAME was still wet behind the ears. I was actively participating on Usenet at the time and recall the hot debate which followed. That wheel was not only invented but well worn in eons before fubar came along. It's an issue of timing - one significantly predated the other. MAD was an original concept evolving from mpglib. BASS is an original unto itself. fubar's decoder is an attempt to copy something already in existance to avoid licensing issues as the author has admitted himself.

It's a personal hangup of mine based on professional *opinion* (yes, I did say the "o" word). I tend to go with originals - not "me toos", especially when those "me toos" are grafted on after the fact. Dithering and soundshaping were most definitely not there when fubar debuted. :)

[hedge]

I gotta say this, I can't see a damn thing wrong with foobar. Its infinitely customizable, loads a total of 20000 songs spread through five playlists in mere seconds (cough cough database!!!!), has the same if not better (due to customizability once again) tagging as qcd, supports more formats in well made plugins that do not get left in a 'meh can't be bothered with this no more' state, has an incredible (alright can be slightly confusing... but it allows you to do just about anything) playlist title display system, and I for one can make out NO difference in audio quality, as I'm sure almost everyone couldn't.

Roj, give it a try, it really is good. We all know your feelings regarding the HA guys, but whilst the foobar developer may have links to them, it is not their project. I for one think it a good idea to at least take a look at foobar's sound engine as a viable alternative.

[Toe]

Gonna use Usenet style quoting - too tired. :)

>>Xing is even older, and therefore, it HAS to be the best! :p

But of course - so does the Model-T... :)

>>Then we need to go with Fraunhofer! :p

...in any color as long as it's black (reference to Model-T).

>>Indeed, I looked over some posts from the fb2k developer on kernel
>>streaming, and I really start to get the feeling that Roj is perhaps confusing
>>the opinion of some of fb2k's zealot users with those of fb2k's dev himself.
>>Maybe I missed something...

That could well be - there was a lot of noise (in the communications definition of the term) surrounding fubar and KS back in the day, particularly from one airheaded mod at 3DSS. I just tossed the works and it was so long ago it doesn't matter. Give a dog a bad name and hang him, y'know...

>>I doubt many of QCD's users would be/are too crazy about foobar's
>>interface (that's probably the #1 gripe anybody would have about it), but
>>you've gotta admit there's some things it does better than pretty much
>>anything else out there. As hedge mentioned, its playlist loading speed is
>>excellent, supports having multiple playlists in tabs, its tagger is quite
>>powerful, in some respects better than QCD's (not to mention that it
>>natively supports APEv2 tags in MP3s), I've been pointing to its ReplayGain
>>support as a model for QCD to follow for ages now, built-in Shibatch
>>SuperEQ (even Paul has said it would be a good project to turn that EQ
>>into a native QCD plugin), 32bit pipeline all the way from decoder through
>>DSP and out to the output plugin.

I don't do the multiple playlist thing so that's lost on me. I typically right-click on a folder in Explorer and do the "play in QCD" thing. Advanced playlist functions and sorting and tabbing and all that as a waste of time as far as I'm concerned. I *never* use a player to tag files - MP3tag does that for me and it's far more versatile than any player tagger. ReplayGain is similarly lost on me - if I want that, I use MP3Gain and manually tailor whatever album extracts I've consigned to a given folder.

In short, all of fubar's "advances" are hindrances where I live. Just gimme good sound, a decent interface and get on with it - that's pretty much it for me.

>>(frankly, most of this is probably moot, I really doubt Paul has any serious
>>plans to change the default decoder...)

I do believe you're right.

[Roj]

I gotta say this, I can't see a damn thing wrong with foobar. Its infinitely customizable, loads a total of 20000 songs spread through five playlists in mere seconds (cough cough database!!!!), has the same if not better (due to customizability once again) tagging as qcd, supports more formats in well made plugins that do not get left in a 'meh can't be bothered with this no more' state, has an incredible (alright can be slightly confusing... but it allows you to do just about anything) playlist title display system, and I for one can make out NO difference in audio quality, as I'm sure almost everyone couldn't.

Roj, give it a try, it really is good. We all know your feelings regarding the HA guys, but whilst the foobar developer may have links to them, it is not their project. I for one think it a good idea to at least take a look at foobar's sound engine as a viable alternative.
I did in the beginning and kept tabs on it all the way to 0.6 when it slipped off my radar as the mundane often does. It promised far more than it delivered and the whole "minimalist" schtick cut zero ice with me. The latter came across (and still does) as someone who just didn't want to be bothered with the niceties. After all, when you promise "sound and slimness over bloat" and don't even deliver the sound let alone the slimness, what other conclusion is a body to come to? As an example, when more features were added as the project progressed, it used more resources than QCD - and that *still* without the skinning and other features it decried while still claiming "minimalism". In short, for me it was and remains more hype than substance.

Now, I'll grant you that the audio quality *has* improved significantly since it's creation. However the shrill claims of its adherents, the religious fervor surrounding it and that retroactive abortion of an interface it uses incline me to take claims regarding it with a HUGE grain of rocksalt. For me it's more trouble than it's worth.

[Roj]

DAMN!!!!!!!!!!!

My sincere apologies to Toe.

I inadvertantly butchered your post.

I'm bushed this morning and accidentally hit Edit instead of Reply, my mod priveleges kicked in allowing me to do this and I saved a completely bastardized version of your post.

I really should NOT do this before my first cup of coffee, especially not after being up until 2:00AM troubleshooting.

Sorry, mon.

[hedge]

As they say, to each his own. Still, if Paul doesn't mind he could at least take a look at the foobar audio implementation, and give us his thoughts, because ultimately hes the one doing the work :evolved: .

(ps. i just came from the best damn gig i've ever been too... I'll just say this, IN FLAMES RULE!!!! (oh and that I'm getting some good ear plugs, every gig i go to it just seems to be getting louder and louder... but hey going by my sig i should like that kinda thing... ) :normal: )

[madjo]

if I understand correctly... Roj wanted to have the default mp3 decoder replaced.. but euhm, isn't this decoder just simply an input plugin, and if that one is changed the decoder is changed?

in other words... if it can be done by plugins, why change the default decoder if it is just a handfull of users that want it changed? :)

please don't slam me to hard ;-) I know little about this...

[jkrzok]

if I understand correctly... Roj wanted to have the default mp3 decoder replaced.. but euhm, isn't this decoder just simply an input plugin, and if that one is changed the decoder is changed?

in other words... if it can be done by plugins, why change the default decoder if it is just a handfull of users that want it changed? :)

If a better sounding mp3 input was available the default should be changed. Why not go for the best if the best is available? If nothing else, it would improve new user's first impression about the overall quality of the product. You never get a second chance to make a first impression. I would imagine that many first time users who thought the player sounded bad would blame the player itself and not the input plugin. Many wouldn't try another plugin, they'd just delete the player.

There's been almost a million downloads of QCD on Download.com. What percentage of those are savvy enough to even know you can change an input plugin? If they're like me they came from Musicmatch or something similar that can't even change plugins. Can WiMP change plugins? Or Real?

Always put your best foot forward. Ultimately this is Paul's decision and I believe that if a better sounding plugin is available he'll change it.

[Toe]

Ultimately this is Paul's decision and I believe that if a better sounding plugin is available he'll change it.
I'd say it as "Ultimately this is Paul's decision and I believe that if a better quality plugin is available and he deems it worth the hassle to switch he'll change it." :)


EDIT: I suppose there's one more clause that has to be added: "if he has TIME to change it" :(

[Roj]

I'd say it as "Ultimately this is Paul's decision and I believe that if a better quality plugin is available and he deems it worth the hassle to switch he'll change it." :)


EDIT: I suppose there's one more clause that has to be added: "if he has TIME to change it" :(
I agree completely on all points - as I said before, fatherhood is VERY time consuming. :)

[brian]

The long and short of it is that a fully functioning MAD input plugin is needed. Nobody here has claimed that anything better than this could be found. It is surely technically possible, as CoolPlayer demonstrates. For my money such a plugin should also become the default, although I'd prefer to see the concept of `default' replaced by `recommended' to take account of QCD's modular structure. In fact, unless I've missed something, when you download, install and use QCD, nowhere does it say anything about defaults.

[hedge]

The long and short of it is that a fully functioning MAD input plugin is needed. Nobody here has claimed that anything better than this could be found. It is surely technically possible, as CoolPlayer demonstrates. For my money such a plugin should also become the default, although I'd prefer to see the concept of `default' replaced by `recommended' to take account of QCD's modular structure. In fact, unless I've missed something, when you download, install and use QCD, nowhere does it say anything about defaults.
The closest we have is Case's effort, which I haven't had any problems with (apart from slow seeking, but thats a requirement for gapless playback as far as i know, and as i never stream I of course haven't encountered these bugs), but others have had problems with streaming and some problems with normal mp3's from time to time.
If this plugin could be polished up a little, I'm sure it would make do quite nicely.

[Rex_Mundi_Incarnit]

...to take account of QCD's modular structure...

Lol i almost read: QCD's molecular structure! :silly:

[Roj]

The long and short of it is that a fully functioning MAD input plugin is needed. Nobody here has claimed that anything better than this could be found. It is surely technically possible, as CoolPlayer demonstrates. For my money such a plugin should also become the default, although I'd prefer to see the concept of `default' replaced by `recommended' to take account of QCD's modular structure. In fact, unless I've missed something, when you download, install and use QCD, nowhere does it say anything about defaults.Man, I *love* the way this discussion hsa turned around. :) This is a *very* good idea but it hinges on one thing, namely a superlative MAD plugin.

To that end, IMHO such a plugin should have as its absolute non-negotiable requirements the following:

- bug free (or as bug free as one can make it - we should ALL be involved in BETA testing this beast)
- the full suite of MAD controls inherent in the Winamp flavor
- seamless / gapless playback
- full streaming support including song title display
- fast seeking
- ironclad MP3 format playback (no "I've got an MP3 that MAD won't play" scenarios)
- ReplayGain (I don't particularly care for this but many others do)
- Equalizer support
- consistent technical support (if something goes wrong with it, it will be fixed in a prompt and timely fashion - this is FAR more important than its position in this list belies)

That's the best I can do after having just woken up - suggestions are welcome.

The reason I nominated BASS is that it fulfills the vast majority of this list already, right down to the support issue, while having audio quality and versatility equivalent to MAD. Either one would be fine for me.

[Aaron]

I agree completely on all points - as I said before, fatherhood is VERY time consuming.

As the bleary-eyed but very proud father of a 1 week old little boy I'd have to agree. I even updated my avatar :)

The list you make for an ideal MAD plugin makes it sound like something worth waiting for.

Interesting point about the consistent technical support. How do you propose this might work? There are two reasons why someone might write software - love or money. As a coder myself the first will get something written but the second helps get it consistently supported.

Cheers,

Aaron

[Roj]

As the bleary-eyed but very proud father of a 1 week old little boy I'd have to agree. I even updated my avatar :)
CONGRATULATIONS!!!! :) :) :)

Interesting point about the consistent technical support. How do you propose this might work? There are two reasons why someone might write software - love or money. As a coder myself the first will get something written but the second helps get it consistently supported.
That's a very valid point and it makes an even better case for BASS. Why? Because as I pointed out, the majority of the work is already done given the advertised feature set and Ian is on the hook for the decoder because he DOES charge for it (that's my corporate persona talking :) ).

[madjo]

If a better sounding mp3 input was available the default should be changed. Why not go for the best if the best is available? If nothing else, it would improve new user's first impression about the overall quality of the product. You never get a second chance to make a first impression. I would imagine that many first time users who thought the player sounded bad would blame the player itself and not the input plugin. Many wouldn't try another plugin, they'd just delete the player.

There's been almost a million downloads of QCD on Download.com. What percentage of those are savvy enough to even know you can change an input plugin? If they're like me they came from Musicmatch or something similar that can't even change plugins. Can WiMP change plugins? Or Real?

Always put your best foot forward. Ultimately this is Paul's decision and I believe that if a better sounding plugin is available he'll change it.

ah ok... Now that I understand it better, I agree with you there :-)
but I must say that I don't really hear the differences :) (not that that says anything... some of you have better musical hearing)

but indeed if there is a better decoder, and the plugin becomes available, it should be the standard. :)

[brian]

I'm all in favour of a developer recouping the investment of their time and expertise that would be needed for ongoing support. But I must say that if a MAD input plugin did everything that was required of it straight out of the box, I wouldn't be too bothered if it turned out to be a one-off production.

[jkrzok]

but I must say that I don't really hear the differences (not that that says anything... some of you have better musical hearing)

You gotta figure many of the differences people hear are just a placebo effect, especially if most people are powering $10 speakers with their integrated soundcard (I doubt this applies to many people in this forum).

That being said, I still think most people would be better off with new speakers or soundcard vs. worrying about a plugin.

[Roj]

You gotta figure many of the differences people hear are just a placebo effect, especially if most people are powering $10 speakers with their integrated soundcard (I doubt this applies to many people in this forum).

That being said, I still think most people would be better off with new speakers or soundcard vs. worrying about a plugin.
I agree. The problem is that when you DO get those new speakers and sound card, a lot of stuff that previously sounded great then sounds like absolute ca ca. :)

[Roj]

I'm all in favour of a developer recouping the investment of their time and expertise that would be needed for ongoing support. But I must say that if a MAD input plugin did everything that was required of it straight out of the box, I wouldn't be too bothered if it turned out to be a one-off production.
Unfortunately MAD is missing a number of the features on my list, notably streaming titles, ReplayGain and my pet peeve seamless / gapless play.

[hedge]

Unfortunately MAD is missing a number of the features on my list, notably streaming titles, ReplayGain and my pet peeve seamless / gapless play.
Case's has both replaygain and seamless/ gapless playback as far as i know....
just gotta get those bugs worked out.
Any luck in contacting him?

[brian]

I trust this won't annoy anyone, but I've been trying out the Winamp mpg123 plugin (SSE2 version, for Pentium 4) to play mp3 streams in QCD via the Winamp input plugins manager. It shows excellent functionality, and I must say it sounds pretty darned good. Of course there is an equivalent QCD plugin, but as already noted by commentators on the plugins page, it unfortunately doesn't work too well. I can't get it to work at all with mp3 streams.

[matty28carter]

Case's has both replaygain and seamless/ gapless playback as far as i know....
just gotta get those bugs worked out.
Any luck in contacting him?

No such luck atm...
:depressed

[Toe]

Huh, I'd never even heard about Case's MAD plugin, I'd only seen shaohao's one. (I don't read the plugin boards very often) He does have the source posted, tho...

http://www.saunalahti.fi/cse/audio.php

[Roj]

I trust this won't annoy anyone, but I've been trying out the Winamp mpg123 plugin (SSE2 version, for Pentium 4) to play mp3 streams in QCD via the Winamp input plugins manager. It shows excellent functionality, and I must say it sounds pretty darned good. Of course there is an equivalent QCD plugin, but as already noted by commentators on the plugins page, it unfortunately doesn't work too well. I can't get it to work at all with mp3 streams.
I'm no fan of mpg123 personally. Too harsh and tinny to my ears compared to the dithering that both MAD and BASS offers.

But that's just me...

[DanTheManPR]

Unfortunately MAD is missing a number of the features on my list, notably streaming titles, ReplayGain and my pet peeve seamless / gapless play.

I'm a little weak on terminology. What is ReplayGain and seamless/gapless play, specificaly?

[Roj]

I'm a little weak on terminology. What is ReplayGain and seamless/gapless play, specificaly?ReplayGain: Making all the songs in a given playlist have the same average volume. It's done on-the-fly.

Seamnless / gapless playback: If a sequence of songs is recorded as a mix (such as those done in a dance club), there will be no slight gap between those song files during playback.

[jkrzok]

For replay gain take a look at http://www.replaygain.org/

For a nice bit of software that does replay gain take a look at http://mp3gain.sourceforge.net/index.php. This adjusts the mp3 file itself losslessly and is reversible. The change even stays if you burn an audio cd of the files. If your mp3s are from multiple sources it's likely they have different volumes. There's nothing nice about a large unnecessary volume change waking you from your nap.

For me the greatest advantage of gapless playback is in live concert recordings. There's nothing more annoying than that little pause in what should be a seamless sound.

[Quake]

I wonder if Paul is checking the forums...

[jkrzok]

I've seen him here not to long ago. He may even be here anonymously. :paranoid:

[Roj]

I've seen him here not to long ago. He may even be here anonymously. :paranoid:
If he is, I'd love to hear some kind of definitive statement from him on the possibility of replacing the default decoder.

[Quake]

This is off-topic but...
Roj, What do you think about the Logitech Z-640 5.1 Speakers?

[Roj]

This is off-topic but...
Roj, What do you think about the Logitech Z-640 5.1 Speakers?
Good for games but not much else.

[matty28carter]

case's mad plug-ins source is available, can't anyone continue where he left off (thats if he isn't continuing...) :scared:

[DanTheManPR]

ReplayGain: Making all the songs in a given playlist have the same average volume. It's done on-the-fly.
That sounds awsome.

[Roj]

That sounds awsome.
Not my cup of tea at all. I personally view it as a hack. If I want that functionality I run a proggie mentioned here called MP3Gain. It does a thorough two pass procedure, analyzing and then losslessly altering the mp3 file. No chintzy potentially sound warping tweaks...

[Toe]

Not my cup of tea at all. I personally view it as a hack. If I want that functionality I run a proggie mentioned here called MP3Gain. It does a thorough two pass procedure, analyzing and then losslessly altering the mp3 file. No chintzy potentially sound warping tweaks...
WTF, you consider a tag that tells the player to slightly adjust its volume control on playback to be a hack? :rolleyes:

Roj, MP3Gain USES ReplayGain to determine what the volume adjustment should be. That "two pass procedure" was written by David Robinson, creator of RG. Really, MP3Gain is very sligtly inferior in its adjustments vs ReplayGain. The variables within the MP3 blocks that MP3Gain changes can only be adjusted in 1.5dB increments. ReplayGain by itself can be adjusted in much finer increments.

MP3Gain is reversible, that is, the MP3 can be restored to more or less its original state after the fact. However, last I checked, the only record of the changes MP3Gain makes is its log file. If for whatever reason you don't have that log file, you're pretty much SOL on being able to undo your changes. (I sincerely doubt you're going to remember the exact adjustment value for every track in your collection.)

Even though MP3Gain can be reversed, it is not truely lossless - by definition any volume adjustment is lossy. If you honestly think you can hear a difference whether the transform is performed at the MP3 decode stage vs after, or for that matter if it's done by a potentiometer, I'm afraid all I can do is laugh.



(hmmm... RG data to control an IC pot... interesting, if not particularly useful or practical...)

[krazyd]

Replaygain can be used on any music format. A major advantage.

[DanTheManPR]

Replaygain can be used on any music format. A major advantage.
exactly. until mp3gain works for ogg and ape, it only serves part of its purpose.

[Roj]

WTF, you consider a tag that tells the player to slightly adjust its volume control on playback to be a hack? :rolleyes:

Roj, MP3Gain USES ReplayGain to determine what the volume adjustment should be. That "two pass procedure" was written by David Robinson, creator of RG. Really, MP3Gain is very sligtly inferior in its adjustments vs ReplayGain. The variables within the MP3 blocks that MP3Gain changes can only be adjusted in 1.5dB increments. ReplayGain by itself can be adjusted in much finer increments.

MP3Gain is reversible, that is, the MP3 can be restored to more or less its original state after the fact. However, last I checked, the only record of the changes MP3Gain makes is its log file. If for whatever reason you don't have that log file, you're pretty much SOL on being able to undo your changes. (I sincerely doubt you're going to remember the exact adjustment value for every track in your collection.)

Even though MP3Gain can be reversed, it is not truely lossless - by definition any volume adjustment is lossy. If you honestly think you can hear a difference whether the transform is performed at the MP3 decode stage vs after, or for that matter if it's done by a potentiometer, I'm afraid all I can do is laugh.



(hmmm... RG data to control an IC pot... interesting, if not particularly useful or practical...)Any technology that "guesses" a level on the fly ain't worth spit to me. That's the same reason I laughed at DBX years ago and tossed it on the rubbish heap where it belonged ("pump pump pump"). The two pass approach is the only correct one as far as I'm concerned. When one records a tape or transfers vinyl to CD, one doesn't "guess" the maximum level on the fly - that's an open invitation to distortion. One goes through the whole song (or whole album) and adjusts the gain so that nothing clips - then one records.

So ReplayGain is more granular - big deal. A granular guess is still just a guess.

I couldn't care less about reversing the process although MP3Gain is reputed to store the information in the ID3 tag attached to the file. Given that my original mp3s are on (several hundred) CDs or DVDs and the altered versions are on my server, I can recall the original MP3 any time I wish. In fact, I *remove* the tags once MP3Gain is finished.

I have to laugh at your statement that volume adjustment is lossy. If the file is not being re-encoded and the content remains unchanged, it's lossless - period. If the perception of dynamic range is altered, that still ain't lossy - that's your hearing. By your "definition", turning up my amp is somehow altering the material on the original CD. Uh HUH.

I'm not claiming I can hear the difference with MP3s but from a simply engineering perspective, something that adjusts on the fly rather than accurately measuring range first and THEN adjusting is just plain *sloppy*.

The only use I have for adjusting the volume in the first place is for compilations, such as the various "miscellaneous" folders on my server which hold several hundred songs apiece or for the 8 hour compilation extract CDs I play in my car and portable. Naturally, anyone who uses that sort of adjustment on a pre-packaged album isn't even worth having the discussion with in the first place.

If you shoose to use ReplayGain with a lossless format (I see the argument there that ReplayGain can be used with more than one format), then more power to you. I didn't spend the money on the Martin-Logans (which is where my lossless downloads are specifically destined) to experience that kind of audio debauchery, thanks very much.

But hey, YMMV of course.

[Toe]

Any technology that "guesses" a level on the fly ain't worth spit to me. That's the same reason I laughed at DBX years ago and tossed it on the rubbish heap where it belonged ("pump pump pump"). The two pass approach is the only correct one as far as I'm concerned. When one records a tape or transfers vinyl to CD, one doesn't "guess" the maximum level on the fly - that's an open invitation to distortion. One goes through the whole song (or whole album) and adjusts the gain so that nothing clips - then one records.

I'm not claiming I can hear the difference with MP3s but from a simply engineering perspective, something that adjusts on the fly rather than accurately measuring range first and THEN adjusting is just plain *sloppy*.
You're seriously, seriously mistaken on how RG is calculated/applied. It doesn't "guess on the fly" any more than MP3Gain does. Did you completely miss my statement that RG and MP3G are 95% identical? The calculations of RG and MP3Gain are 100% identical. You run your MP3s through ReplayGain, it calculates the needed volume change using the ReplayGain algorithm. You run your MP3s through MP3Gain, it calculates the needed volume change using that exact same ReplayGain algorithm! Both are done as preprocessing to the files, rather than as the track is played. The 5% that's different between the two is how they store the necessary adjustment. One takes the value and stores it in a tag inside the MP3 file so that the player can adjust the volume at playback, the other changes a volume multiplier inside every 'block' of the MP3's data. The end result is the same, with the (fairly minor) exception of MP3Gain only being able to change that multiplier in 1.5dB increments.

I couldn't care less about reversing the process although MP3Gain is reputed to store the information in the ID3 tag attached to the file. Given that my original mp3s are on (several hundred) CDs or DVDs and the altered versions are on my server, I can recall the original MP3 any time I wish. In fact, I *remove* the tags once MP3Gain is finished.
I just looked this up, actually. You're right, MP3Gain does also store the undo information in a tag. An APEv2 tag, in fact. Can you guess where ReplayGain data is stored? An APEv2 tag! See my request for native APEv2 tag support in QCD, it's in the frequent requests thread. It's a very nice format, much nicer than either version of ID3, and goes great with RG support in *any* form. :)

I have to laugh at your statement that volume adjustment is lossy. If the file is not being re-encoded and the content remains unchanged, it's lossless - period. If the perception of dynamic range is altered, that still ain't lossy - that's your hearing. By your "definition", turning up my amp is somehow altering the material on the original CD. Uh HUH.
...

I'm sorry, Roj, but volume adjustment is pretty much impossible without distorting the original signal. In digital, you *could* get a lossless transformation by multiplying every sample by the same number, but you've still got problems like rounding the result to the nearest 16bit value. In analog, your signal is being degraded by every bit of copper or silicon it travels through, so where do you even draw the line? That's just the way physics in our universe works. And that's leaving out real-world problems for both like clipping...

The only use I have for adjusting the volume in the first place is for compilations, such as the various "miscellaneous" folders on my server which hold several hundred songs apiece or for the 8 hour compilation extract CDs I play in my car and portable. Naturally, anyone who uses that sort of adjustment on a pre-packaged album isn't even worth having the discussion with in the first place.

If you shoose to use ReplayGain with a lossless format (I see the argument there that ReplayGain can be used with more than one format), then more power to you. I didn't spend the money on the Martin-Logans (which is where my lossless downloads are specifically destined) to experience that kind of audio debauchery, thanks very much.

But hey, YMMV of course.
If all I were doing was listening to very well recorded classical pieces through good speakers, hell no I wouldn't ReplayGain them! RG is intended to help combat the loudness race and clipping on modern CDs. You familiar with the recent dual-layer SACD/CD release of Pink Floyd's Dark Side of the Moon (http://www.stereophile.com/news/11649/)? The SACD layer had the 'normal' levels applied to it, with full dynamic range, while the CD layer on the same disc had been severly compressed by the mixer, resulting in much lower dynamic range than CD is capable of. (Earlier CD releases of DSotM weren't nearly as over-compressed, and thus have much better dynamic range than this 'remaster'.) That's the kind of BS that RG (again, in any form) can at least somewhat alleviate.

[Roj]

Without reading beyond the first couple of sentences:

I didn't miss your statements - I just have trouble buying into them because of the following physical inconsistencies:

In one case, the software takes the time to quickly analyze the entire mp3 and establish a profile of it. This takes a couple of seconds *at best*. A second pass then adjusts the file. That takes a couple of seconds more, which could arguably be just I/O so that could be discounted but you still have to take into consideration the very finite original amount time taken for the first pass.

In the other case, the file just plays with no time lag of a couple seconds. And I'm supposed to believe the same process happens in real time? How else can that be categorized than on the fly with no complete end-to-end file analysis?

I'm looking at the hard evidence of my eyes and fancy rhetoric aside, the math doesn't add up.

[Toe]

Without reading beyond the first couple of sentences:
Read first, then run your mouth. The six hours between your original post and the last time you edited it is plenty of time to read my post, instead of plugging your ears and singing "La la la la la la I can't hear you!" yet again. You're supposed to be a moderator here, aren't you? Ever considered acting like one?

As for your "physical inconsistencies" bit, I've already said several times that RG doen't do its calculations in realtime. MP3Gain directly states that it uses ReplayGain (http://mp3gain.sourceforge.net/faq.php#peak). If you're not even willing to believe the program's own author, I just ran an album through ReplayGain, and then ran another copy of the same album through MP3Gain. RG took 1m 40s to finish. MP3Gain took 1m 52s to finish. That difference sounds about right to me, the calculations are the same and take up the bulk of the time, but where RG is just saving a small tag to each file while MP3G is modifying MP3 frames themselves. The math adds up quite plainly.

[jkrzok]

Do I dare enter this discussion? But I'm confused so here goes.

I thought that if a player/decoder supported ReplayGain it did so internally, that there was no need for a stand alone app that analyzed the files before they were played. I thought ReplayGain worked WHILE the files were being played, on the fly. So if I understand Toe there actually is an app you run a file through before you play it, correct? Isn't there a a plugin that does that? What are you using to run a file through ReplayGain?

I run MP3Gain for my old time radio files so I know roughly how long it takes. It never made sense to me that ReplayGain could work in real time and be effective. But if you change tag info before they're played I understand. I think.

Be gentle. :scared:

[Toe]

I thought that if a player/decoder supported ReplayGain it did so internally, that there was no need for a stand alone app that analyzed the files before they were played.
Your confusion here essentially stems from the fact that there's two parts to a ReplayGain system - one part that scans the file and determines the volume adjustment needed and then saves the result in a tag, and another part that reads that tag and actually applies the volume adjustment at playback time. The scanning and tagging part doesn't have to be done by the player, but it can be.

It's just like tag editors, really. All that the player really has to be able to do is read the tag and display it. But a lot of players also include the ability to edit tags, plus there's a lot of standalone tag editors out there that aren't tied to any particular player. Likewise, all that a player has to do for ReplayGain is read the tag apply the adjustment at the time the track is played. (Or ignore it alltogether, if RG is shut off in the player's preferences.)

If Paul wanted to, he could provide 'read only' support for ReplayGain, meaning that QCD would be able to read the RG tag and apply the volume change. We would be left to use an external app to scan our collections and tag them. Then somewhere down the road, he might also add the ability to scan/tag the files, for 'full' support.

[Roj]

Read first, then run your mouth. The six hours between your original post and the last time you edited it is plenty of time to read my postBefore I answer, let me state one thing:

When you live in my skin, then you will be qualified to tell me how to spend my time.

I'm not under any obligation to read anything you write. How I spend my online time (or time in general) is my choice and quite beyond the dictates of anyone but myself on what I choose to do with it. You may have noticed that my interest in participating in online forums is rather limited as of late. What I'm reading here is certainly not conducive to any alteration of that general trend.

That being said, I stand corrected on the methodology. I do tend to avoid what i view as miscellaneous unnecessary tags. As I stated before, once MP3Gain is done its thing I remove the tags that allow reversal. Call it paranoia on my part since many of my compilations end up in hardware devices other than my computer and all firmware is not created equal in my experience.

I reiterate my stance on using the technique with lossless but concede that not everyone has the tastes or equipment that compel them to be the purist that I prefer to be.

That about expends all the time I feel like spending on the subject (and boards in general) today. Back to listening to Bauhaus and casual surfing before dinner and whichever DVD movie was in my download queue for the day... :)

[Toe]

Before I answer, let me state one thing:

When you live in my skin, then you will be qualified to tell me how to spend my time.
I don't live in your skin, but I do know that you're a moderator on this forum. Being a moderator involves a certain time commitment, and I would hope that you show the users of this forum some courtesy and professionalisim, set a good example. I haven't seen you doing these things here.

Mind you that I'm a forum admin myself.

That being said, I tend to avoid miscellaneous unnecessary tags. As I stated before, once MP3Gain is done its thing I remove the tags that allow reversal. Call it paranoia on my part since many of my compilations end up in hardware devices other than my computer.
Yeah, that's one thing that's nice about APEv2 tags, specificly designed not to fuck up player compatibility, and does a very nice job of it. But if you prefer the paranoia route, so be it.

I stand corrected on the methodology.
There, now that wasn't so hard, was it? :)

I reiterate my stance on using the technique with lossless but concede that not everyone has the tastes or equipment that compel them to be the purist that I prefer to be.
Again, so be it, but if you're really a purist, WTF are you doing using MP3 in the first place?


EDIT:
That about expends all the time I feel like spending on the subject (and boards in general) today.
Funny, then what are you doing coming back and editing your posts more than a half hour later? :P

[jkrzok]

If Paul wanted to, he could provide 'read only' support for ReplayGain, meaning that QCD would be able to read the RG tag and apply the volume change. We would be left to use an external app to scan our collections and tag them. Then somewhere down the road, he might also add the ability to scan/tag the files, for 'full' support.

When I look at the options of the new version of MAD ReplayGain looks like it is supported. Is it enough for the plugin to support it or does the player itself need to have that functionality built in?

And what are you using to do your ReplayGain analysis if not MP3Gain?

[Roj]

I don't live in your skin, but I do know that you're a moderator on this forum. Being a moderator involves a certain time commitment, and I would hope that you show the users of this forum some courtesy and professionalisim, set a good example. I haven't seen you doing these things here.
First off, you're wrong about the time commitment in this case and secondly, I never have been able to suffer fools gladly. I also happen to be of the mind that political correctness is a waste of air so I call a spade a spade. That may occasionally offend people. Too bad. I don't babysit - if they need that, I suggest that they get off the net and hide under their bed.

I've never treated anyone in any fashion that I wouldn't expect to be treated myself. Since I don't suffer fools gladly, I don't expect anyone to put up with me if I'm behaving like one. YMMV. I've also been known to back down when I'm wrong. Of course, first whoever is trying to do that will have to prove it beyond conjecture and empty rhetoric.

Mind you that I'm a forum admin myself.EEK! :)

Yeah, that's one thing that's nice about APEv2 tags, specificly designed not to fuck up player compatibility, and does a very nice job of it. But if you prefer the paranoia route, so be it.The first thing I do when I install a new version of MP3Tag is go to the mp3 section and have it delete all APE tags. Similarly, I don't have it put ID3 tags on APE files.

There, now that wasn't so hard, was it? Like I said, I call a spade a spade. :)

Again, so be it, but if you're really a purist, WTF are you doing using MP3 in the first place?MP3 is portable, it's easy to amass and it gives me exposure to vast amounts of music that I would otherwise never come in contact with. In a former life I was a club DJ. I had a vast collection of vinyl (I still have a few thousand pieces). Then I discovered MP3. It was FREE and a welcome change after the avaricious thievery of the record companies. I OD'd massively - at this point I have more music than I could ever hope to listen to in this lifetime (we're talking more than a few hundred CDs full of MP3 albums). I play them on the bus, I play them in my SUV, I play them on the plane, I take them with me to parties and BBQs...

...and if I like the album enough I either download it lossless or (Horrors and shock!) actually BUY it.

Besides, I like filet mignon but I also occasionally eat at Quiznos (YUM!).

/rant on

I have the peculiar mindset that says I've paid the record companies far more than enough in this lifetime and that based on my experience with their voracious greed (having seen vinyl prices rise by 1000% over a three year period), I'm perfectly entitled to whatever music I want for free for the rest of my (un)natural life and I still won't break even for what they've ripped me off. :)

/rant off

Funny, then what are you doing coming back and editing your posts more than a half hour later? :PIt's simple. I despise what I consider to be loose ends. I also pride myself on what I write and if I can formulate a better way of expressing myself (it happens completely in background and I'm not even particularly conscious of it occurring) then I come back and refine. The intent is not to change what's there – it's just the expression of the continual thought process. As an example, I will write an article (I occasionally do articles on the web when I feel I have something to say) in a single day. It won't reach the web for three more weeks or even longer. I won't work on it every day either - I'll just let it sit. When the refinements come (alway unbidden), I'll make them. At some point I'll get a mental flag raised that says "you're done Roj". That's when the article makes it out the door. Boards make that process more difficult - they're instantaneous. However, the same mindset drives my participation there. Don't be surprised if this post gets edited again. :) Right now it's sitting in OpenOffice being composed.

[Todd The Kiwi]

Then I discovered MP3. It was FREE and a welcome change after the avaricious thievery of the record companies
yeah amen to that man
I like filet mignon but I also occasionally eat at Quiznos (YUM!)
i was going to say something about mincing words...:carrot:
hang on,i just did :cross-eye

p.s what does "YMMV" mean?

[hedge]

p.s what does "YMMV" mean?
"Your mileage may vary" - Good ol' google :ditsy:

[Qaz]

First off, you're wrong about the time commitment in this case and secondly, I never have been able to suffer fools gladly. I also happen to be of the mind that political correctness is a waste of air so I call a spade a spade. That may occasionally offend people. Too bad. I don't babysit - if they need that, I suggest that they get off the net and hide under their bed.
But commenting text that you haven't bothered to read...not really a way to do it.

It's simple. I despise what I consider to be loose ends. I also pride myself on what I write and if I can formulate a better way of expressing myself (it happens completely in background and I'm not even particularly conscious of it occurring) then I come back and refine. The intent is not to change what's there – it's just the expression of the continual thought process. As an example, I will write an article (I occasionally do articles on the web when I feel I have something to say) in a single day. It won't reach the web for three more weeks or even longer. I won't work on it every day either - I'll just let it sit. When the refinements come (alway unbidden), I'll make them. At some point I'll get a mental flag raised that says "you're done Roj". That's when the article makes it out the door. Boards make that process more difficult - they're instantaneous. However, the same mindset drives my participation there. Don't be surprised if this post gets edited again. :) Right now it's sitting in OpenOffice being composed.
So you don't publish articles until you think they're finished. That's quite allright. Why not the same behaviour with forum posts?

But all this is getting past the point. Could you clearly say why you're so strongly against ReplayGain? (Main reason RG sounds good to me is that it supports other files than MP3s too, otherwise MP3Gain would be perfectly good for me.)

[brian]

Not much gain from the replay of arguments in this thread...

[Roj]

But commenting text that you haven't bothered to read...not really a way to do it.
True enough, but on occasion it gets wearisome wading through all the verbiage, especially when it appears that verbiage is off base. I do err in that regard sometimes.

So you don't publish articles until you think they're finished. That's quite allright. Why not the same behaviour with forum posts?
You want that I should take weeks to answer a post? Impractical at best.

But all this is getting past the point. Could you clearly say why you're so strongly against ReplayGain? (Main reason RG sounds good to me is that it supports other files than MP3s too, otherwise MP3Gain would be perfectly good for me.)
I harbored a misconception on the way RG worked based on when I first used it. It appeared to me that it did normalizing on the fly which is definitely NOT the way to go. Now that it has been explained to me that it actually processes the entire file first and then plays back, I've conceded that it has its place.

[Roj]

Not much gain from the replay of arguments in this thread...
That is the tidiest and most apropos pun I've seen all week! :)

Kudos!

[kode54]

I investigated myself and found that neither BASS nor XMPlay import system intstalled DirectShow or ACM MP3 decoders, by removing all possible filters and modules and verifying that known ACM/DirectShow players fail to play MP3 files. Regardless of these libraries, both players continue to play both MP3 and MO3 files.

Conclusion: BASS and XMPlay have internal MP3 decoding, as well as Ogg Vorbis. Anybody is welcome to compile a simple application which feeds a MP3 into BASS with the floating point output flag, and then verify that the floating point output is indeed greater than 16 bit precision. If it is not, then there is no point to replacing QCD's default decoder with BASS.

This 16-bit math may be different from 16-bit output precision, but I doubt it matters at all in the case of BASS or XMPlay, as I suspect neither of them perform more than simple truncation/rounding and clipping.

As for the porting difficulties and winelib, I think it has something to do with the wonderful Windows-only DirectShow DMO audio filters, an extension we can all thank Olivier Lapicque for adding to the IT format. Spoony bard.

[Roj]

I investigated myself and found that neither BASS nor XMPlay import system intstalled DirectShow or ACM MP3 decoders, by removing all possible filters and modules and verifying that known ACM/DirectShow players fail to play MP3 files. Regardless of these libraries, both players continue to play both MP3 and MO3 files.

Conclusion: BASS and XMPlay have internal MP3 decoding, as well as Ogg Vorbis. Anybody is welcome to compile a simple application which feeds a MP3 into BASS with the floating point output flag, and then verify that the floating point output is indeed greater than 16 bit precision. If it is not, then there is no point to replacing QCD's default decoder with BASS.

This 16-bit math may be different from 16-bit output precision, but I doubt it matters at all in the case of BASS or XMPlay, as I suspect neither of them perform more than simple truncation/rounding and clipping.

As for the porting difficulties and winelib, I think it has something to do with the wonderful Windows-only DirectShow DMO audio filters, an extension we can all thank Olivier Lapicque for adding to the IT format. Spoony bard.
Thanks very much for the insightful post. :)

[Dystopian]

I'm no fan of mpg123 personally. Too harsh and tinny to my ears compared to the dithering that both MAD and BASS offers.

Wow... you must have an amazing hearing. I also prefer MAD for its gapless capabilities, but have never heard a difference.
Have you bothered ABXing QCD's default decoder (or mpglib/fhg) against MAD?
I've found a page called the MAD challenge (http://ff123.net/madchallenge.html), have you done something similiar?

In the end you can't really know if QCD's decoder needs replacement, if you haven't ABXed it (gapless playback is a good reason, but could be implemented for sure).

I'd appretiate if someone with a hearing like yours could prepare an ABX-based comparisons of all major MP3 decoders. Someone at Hydrogenaudio tried that some time ago, but didn't come up with any useful results (in fact he didn't like the sound of MAD's dithering) and didn't include QCD: Full Results (http://www.foobar2000.net/mp3decoder/conclusion.htm)

dystopian

P.S.: Could you explain what 'soundshaping' does? I understand the concept of dithering and noiseshaping, but what is the effect of soundshaping?

[Roj]

Wow... you must have an amazing hearing. Actually I do. When I was younger I could hear dog whistles) they were a very annoying thin sound at the edge of my awareness). Now that I'm older, I'm told that my hearing is still well above the level normal for my age of 43 (I was tested a year or so ago along with a buddy of mine).

I also prefer MAD for its gapless capabilities, but have never heard a difference.Native MAD has no gapless capabilities. That's a Case innovation (bless him). :)

Have you bothered ABXing QCD's default decoder (or mpglib/fhg) against MAD?No - I just listen. I place zero faith in ABX. I certainly don't do it in an audio showroom so I won't be doing it on a computer any time soon. As far as I'm concerned it's artifical and flawed. However, if I can't hear a difference (and I can't in all good conscience say that I do with MAD, BASS and the latest fubar with dithering and soundshaping turned on in all three), then I say so. My testing regimen involves playing the same snippet of a given mp3 that I encoded in the two players I'm evaluating at my best approximation of the same volume level. I listen for subtle nuances in decay, how instruments resonate and the like. I repeat this as often as I want until I'm comfortable with the results. If I can tell a difference, I note it. If I can't, I note that too. This is quite analagous to how I evaluate audio gear.

I've found a page called the MAD challenge (http://ff123.net/madchallenge.html), have you done something similiar?I remember that page from way back when. At least one of the folks putting that page together hung out on Usenet when I actively participated in that medium (instead of just leeching :) ). There was a long and controversial discussion over it. I tried MAD on my own terms and found it to be different than the other decoders available at the time.

In the end you can't really know if QCD's decoder needs replacement, if you haven't ABXed it (gapless playback is a good reason, but could be implemented for sure).Sure you can. I don't need to ABX *anything* to test its audio quality. All I have to do is *listen* to it. That simple test was around eons before anyone thought to drag psuedo-science into the equation.

I'd appretiate if someone with a hearing like yours could prepare an ABX-based comparisons of all major MP3 decoders.Good luck to them. I can guarantee you it won't be me... :) My ears have and continue to serve me well with the system I have used for a very long time.

Someone at Hydrogenaudio (http://www.hydrogenaudio.org/forums/index.php?showtopic=17728) tried that some time ago, but didn't come up with any useful resultsFew things on Hydrogen Audio can be classed as "useful results" in my experience. I've lurked there, read, and left shaking my head. My best description of that place is the name of a play by Shakespeare:

Much Ado About Nothing.

Yes, I know I fly in the face of a lot of current audio "religion" but that's pretty much all I consider a lot of it to be. I learned quite early never to put any faith in religion - too much of it is constructed, usually to suit someone. :)

P.S.: Could you explain what 'soundshaping' does? I understand the concept of dithering and noiseshaping, but what is the effect of soundshaping?The best explanation I've seen of both I've lifted from the XMPlay forum:

Dither

Dithering is simply adding a low level white noise to the sound before it's converted (from floating-point) to the output resolution (8/16/24/32-bit). Adding noise may seem silly, but it actually effectively increases the resolution because it allows very low level signals to appear in the output that would not do otherwise (eg. they could have been rounded down to 0) - even though the noise level is higher than the low level signal, the brain can pick out the signal.

Noise Shaping

Noise shaping moves the noise generated by dithering to frequencies that the human ear is less sensitive to, so that it becomes less audible.

Noise shaping is the same as soundshaping as far as I know unless someone has additional information to offer.

Cheers!

[Dystopian]

I place zero faith in ABX.[...]
I don't understand.
If you can't even prove that you're hearing a difference, how can you dare judging which decoder (and/or dithering/noiseshaping method) sounds better?

That doesn't make any sense to me.

dystopian

P.S.: Thanks for your explanation. I've never seen the term 'soundshaping' before, so I wasn't sure if it was some kind of DSP...

[brian]

Golly gosh - I hope that the longstanding and irresoluble Toe-Roj subjectivity/objectivity debate isn't about to be `replayed' as a Dystopian-Roj debate...

[Roj]

I don't understand.
If you can't even proof that you're hearing a difference, how can you dare judging which decoder (and/or dithering/noiseshaping method) sounds better?

That doesn't make any sense to me.
I have the only proof anyone ever needs that something sounds the way it sounds:

My ears.

All else is irrelevant - I have no need to use any "test" (or what passes for one to some) other than that.

If you think about it, neither do you.

What are you going to believe? Are you going to go with what someone says *should* be valid because it conforms to a methodology *they* invented or will you accept the evidence of *your own ears* garnered by doing what comes naturally (and is essentially what music is all about), namely...

...listening?

For me, the answer has and will always make more sense to me.

YMMV.

[Roj]

Golly gosh - I hope that the longstanding and irresoluble Toe-Roj subjectivity/objectivity debate isn't about to be `replayed' as a Dystopian-Roj debate...
Nah.

I leave the tests to those who need the crutch. One iteration of that waste of time is enough. :)

[Dystopian]

What are you going to believe? Are you going to go with what someone says *should* be valid because it conforms to a methodology *they* invented or will you accept the evidence of *your own ears* garnered by doing what comes naturally (and is essentially what music is all about), namely...

In philosophy this is called 'Naive Realism', in science just 'uninterersting'.
But since you don't seem to be interested in putting some value behind your claims, I'll just leave it to others to draw their own conclusions regarding your dialectic.

dystopian

[Roj]

In philosophy this is called 'Naive Realism', in science just 'uninterersting'.
But since you don't seem to be interested in putting some value behind your claims, I'll just leave it to others to draw their own conclusions regarding your dialectic.

dystopianI never took philosophy. :)

I am interested in my own senses and what they tell me.

I always wholeheartedly encourage others to do the same and to form their own opions, especially in audio.

That way, they can judge for themselves in an area fraught with too many variables and too much subjectivity to ever be limited to one simple testing criterion, never what the folks with electronically powered beanies tell you. :)

I never try and make square pegs fit in round holes for the sake of tidiness. I prefer to simply observe the holes for what they are.

The bottom line is that people should *listen* in whatever environment and using whatever method they feel comfortable with - period.

That's the only true test.

[kode54]

Whoops. As a minor correction to my previous post, I forgot that XMPlay supports dithering and noise shaping as separate options. I can only guess that the noise shaping option only applies to dithering, as I cannot imagine how noise shaping would be applied to rounding and truncation, other than with some form of error diffusion dithering.

One of the most basic forms of dithering, error diffusion, is where one sample (or pixel in images) is truncated or rounded to the nearest possible value in the target resolution, then the error is distributed to the next sample. This, of course, introduces some noise into the resulting signal. The idea behind noise shaping is to limit how much of this noise falls within the audible frequency range. Or at least, that is how I understand it to be.

[rorythedog]

At one end of my living room I've got £3000 worth of CDplayer, amp and speakers. At the other I have a couple of beige boxes and a few wee satellite speakers. Frankly, the idea that anything emanating from the latter could, in any way, be described as "lossless" is preposterous.

The original CD's aren't lossless.

Bugger. Looks like I've swallowed the same dictionary as you guys.

Anyone want to come over to mine's and listen to some tunes in a refreshingly chilled and smoke-filled environment? Feel free.

[Quake]

I'm a little confused by this... but I still have the same question...

WHERE THE HELL IS PAUL!

ehem...

[jkrzok]

In Toronto somewhere, mourning the loss of the NHL season, most likely.

[t22design]

I agree with Roj about the listening thing.

As someone who loves music, and has a resonably decent PC audio setup (ie not onboard sound, and a resonable amp and speakers), I have to say that you should let your ears be the judge of what sounds better.

As for changing the decoder - does that mean I'll have to reencode all my CD's if gapless play happens?

:confused:

[Roj]

As for changing the decoder - does that mean I'll have to reencode all my CD's if gapless play happens?

:confused:
Not at all - it does hopefully mean that they will sound better in playback though because a better decoder and soundshaping ensemble will be used. :)

[Toe]

I agree with Roj about the listening thing.

As someone who loves music, and has a resonably decent PC audio setup (ie not onboard sound, and a resonable amp and speakers), I have to say that you should let your ears be the judge of what sounds better.
An ABX test is using your ears, listening to various tracks to see if YOU can actually hear a difference. A 'track' in this case can be a short sample, a full song, or an entire album.

You listen to track A (or in this case, a file played back through decoder A), then to track B (same file played through a different decoder), and then listen to track X. Track X is really either track A or B, but you don't know which one it is. Your job is to determine which one it is. That's it. You don't even have to judge one as 'better' than the other. You can take the test pretty much any way you like, through a pricey soundcard/speakers/etc, through a nice pair of headphones, sitting in front of your computer, sitting in your favorite easychair, whatever you think will help you listen the most critically.

Repeat the test several times (it doesn't really have to be done in one session), and tally up how many times you correctly guessed which one X is. If that number is significantly higher than what would be obtained through pure guessing (ie significantly higher than 50% accuracy), then it's safe to say that you can indeed hear a difference.

Surely, with this spiffy decoder/encoder/amp/whatever with its amazing <insert jargon you probably know nothing about here>, you should at least be able to at least judge whether or not there's a difference between the two, shouldn't you?

In other words, it's putting your money where your mouth is.

[Toe]

For more information on just how strong the placebo effect can be, I highly reccomend the episode of Penn & Teller's Bullshit (http://www.sho.com/site/ptbs/topics.do?topic=bw) about the supposed 'superiority' of bottled water.

They went to New York and conducted a taste test of tap water vs. bottled water. They found that about 70% of them actually prefered tap water.

Then they went to a fancy restaurant in Los Angeles and were selling patrons $8 bottles of L'eau du Robinet (French for "faucet water") and Agua de Culo (Spanish for "ass water"). The patrons gush about how all the different kinds taste, and proclaimed them to be far superior to tap water. Every last bottle was filled from a garden hose out back.

[brian]

Toe - I think there's a flaw in the way you try to apply the placebo concept to audio tests. The point about a placebo is that a person experiences item X as doing them more good than item Y because they believe X to have an active ingredient, whereas in fact it may be the inert one of the pair. In other words, they experience what they expect to experience, based on their beliefs. But if a person listens to the same audio track through various different decoders, they don't necessarily believe or expect anything one way or the other. I'm not challenging the test you describe, just the use of the term `placebo' to pinpoint what's going on here.

[Todd The Kiwi]

crikey,what's the story?

"We at Showtime Online express our apologies; however, these pages are intended for access only from within the United States" :ermm:
bullshit indeed ha ha ha

[Roj]

An ABX test is using your earsOR...

...you could skip the over-complicated psychodrama and just do what people have been doing for a very long time before this psuedo-science came up and just....

...LISTEN.

At appropriately self-tuned matchng volume levels, of course. :)

It produces the same results and has the added benefit of driving the control freaks and pop-scientists totally BONKERS with frustration because it's something they can't control / rationalize.

In the immortal words of one G'Kar:

"I Like it!"

:) :) :)

(A real B5 fan will recall and understand the double-barbed inference of the above)

Don't get me wrong - science definitely has its place in audio. However, do note that people tuned pianos (and many another instrument) for hundreds of years By Ear as are many high-end audio speakers today. The overzealous psuedo-scientists who scream "objectivity" at every turn tend to lose sight of that salient point.

Now, back to something far more relevant than said "tests", namely downloading and auditioning Le Cafe Abstrait Volumes 1 thru 4 in FLAC format before I leave for work...

[nwegner]

Can I just ask that the encoder (optionally) encode the RG tags and/or do the MP3G preprocessing? My portable MP3 player doesn't know about RG, but I still want clipping prevention for it, even if I have to also use RG to get more precision wherever RG is supported.

[lump1]

I followed this thread and registered because of the urgent need for someone here to say this:

Roj, you're being a baby. It's time for you to stop.

Everybody else seems too polite to stop feeding this troll, but they probably should. Since there is a plugin that sounds superior "to your ears" then use it Roj. If you want to make it the player default because you think it sounds better, conduct a double-blind test. Until you do, all you manage to say is "do X because I want X done". You've been repeating what amounts to this while pounding the table for 8 screens now. I call that "being a baby," though I suppose an adolescent might behave the same way. In any case you won't solve this with more table-pounding.

Produce evidence or back down.

[Todd The Kiwi]

g'day
that's some first post you've made there
you're aware that roj is a moderator, right?
i'd be expecting some flames if i was you...
not that we want that to happen aye ;)

[Qaz]

you're aware that roj is a moderator, right?
Why should that make a difference?

[brian]

May I just reiterate that there is no such thing as a `default' mp3 decoder in QCD - its architecture is all based on optional plugins. If fully developed MAD and BASS plugins became available, the current XAudio plugin could and should continue to be available too, for the benefit of those who prefer it.

[Roj]

I followed this thread and registered because of the urgent need for someone here to say this:

Roj, you're being a baby. It's time for you to stop.

Everybody else seems too polite to stop feeding this troll, but they probably should. Since there is a plugin that sounds superior "to your ears" then use it Roj. If you want to make it the player default because you think it sounds better, conduct a double-blind test. Until you do, all you manage to say is "do X because I want X done". You've been repeating what amounts to this while pounding the table for 8 screens now. I call that "being a baby," though I suppose an adolescent might behave the same way. In any case you won't solve this with more table-pounding.

Produce evidence or back down.I'll tell you what:

When you or anyone else who champions double blind testing can actually *prove* beyond a shadow of a doubt that it has any *real* advantage over what people have been doing for longer than you or I have been alive by simply listening, I'll concede that you have a point.

Until then, it goes where most "religion" goes in my life:

File 13.

You see, that's why the debate continues, not just here but anywhere else in audio circles.

You can't.

'nuff said.

See? No flames necessary, unlike those from our new friend here. :)

[Roj]

Why should that make a difference?It doesn't. :)

If you haven't noticed, I've never thrown my weight around as a mod (and never will). What would that prove?

[brian]

This isn't a debate, for the simple reason that there is no common ground regarding what kind of evidence would be relevant to settling the matter.

By the way, a moderator is precisely someone who looks for common ground and doesn't take sides one way or the other. So if Roj is a moderator, I'm Rush Limbaugh.

[Roj]

This isn't a debate, for the simple reason that there is no common ground regarding what kind of evidence would be relevant to settling the matter.

By the way, a moderator is precisely someone who looks for common ground and doesn't take sides one way or the other. So if Roj is a moderator, I'm Rush Limbaugh.Pleased to meet you, Rush. :)

Or not (I can't stand Mr. Limbaugh).

Actually, from where I'm sitting, a moderator is one who on occasion (the rarer those occasions the better) keeps things from getting out of hand and in general stays out of the way as far as *policing* the discussion is concerned. Think of that person in terms of being more of a bartender with the 12-guage under the counter which on occasion he pulls out and fires a warning shot into the air when the patrons are threatening to out-and-out tear up the bar. Your definition is more applicable to someone who's a mediator and that's one job I'd never want. It would prevent that person from ever having an opinion or defending it.

[brian]

Think of that person in terms of being more of a bartender with the 12-guage under the counter which on occasion he pulls out and fires a warning shot into the air when the patrons are threatening to out-and-out tear up the bar.
You mean like Guinan tending bar in `Tent Forward' in Star Trek: The Next Generation?

[Roj]

You mean like Guinan tending bar in `Tent Forward' in Star Trek: The Next Generation?
Great analogy!

Man, she had the juice - I'd never have messed with her. :)

[brian]

She kept a Kardassian vaporizer (or whatever) under the bar and wielded it majestically in one episode when some of the crew ran riot in the bar-room (later found to be due to dream-deprivation). Of course she was about 5000 years old so she'd had plenty of time to learn a trick or two.

[Qaz]

It doesn't. :)

If you haven't noticed, I've never thrown my weight around as a mod (and never will). What would that prove?
Didn't mean to imply that you would somehow use the position wrong. I was just wondering why the fact was even brought up. :)

[Roj]

Didn't mean to imply that you would somehow use the position wrong. I was just wondering why the fact was even brought up. :)
Kewl.

And if i ever have a momentary lapse of reason and actually do so, I sincerely hope one of y'all will let me know in short order.

[Roj]

She kept a Kardassian vaporizer (or whatever) under the bar and wielded it majestically in one episode when some of the crew ran riot in the bar-room (later found to be due to dream-deprivation). Of course she was about 5000 years old so she'd had plenty of time to learn a trick or two.
She and Q were the two reasons I watched that series.

Especially Q... :)

[brian]

But did you notice how in the episode where she and the supposedly omnipotent Q met up (evidently not for the first time), he was very wary of her?

[Roj]

But did you notice how in the episode where she and the supposedly omnipotent Q met up (evidently not for the first time), he was very wary of her?
Oh yeah - that's the one and only time I've seen Q back up for anyone. She made him *nervous*.

[carbonize]

Anyway back to the topic at hand.

I was about to start a thread about this then saw this one. I'm not 100% sure but couldn't they get around the GPL restrictions of using MAD by making the MP3 decoder a dll and open sourcing the dll? I do have to agree that the BASS decoder is an excellent decoder. XMplay (http://www.un4seen.com) has excellent sound quality and another player that uses BASS is VU Player (http://www.vuplayer.com) which is a very nice, if somewhat basic, player.

[carbonize]

Well Paul has hinted that the new build is approaching a release, even if only beta. So I just thought, with all the requests for new MP3 decoding and the mad & bass plugins, has he changed the decoding engine used?

[brian]

Somehow I doubt it - he made no mention of decoders when he gave us a rundown of planned changes.