Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bluey]

Ok so I encoded 2 tracks, a chemical brothers and kate bush, since I can do this, I can
play 2 zones together on the same card using different channels of my mixer (I have 32 channels its a pulsar card), this way I can verify that output levels are the same.  It is important to use the same card since the same DA convertor.

So with both playing with the same internal and external volume settings and no DSP settings, I played both files with a 5 second gap between each, and muted and switched between APE and MP3 ..... I couldnt tell any difference whatsoever.  I have them connected to tannoy 800 monitor speakers.

I think its all some ultra audiophile snobbery thats being propagated around the internet about "lossless" formats.  Sure Ape is probably good for archiving purposes since it is format independent to be converted in the future to MP3, WMV, ACC or whatever.

Maybe its the source material I used, but can anybody else tell a difference, and remember to justify if your test setup was flawed in the past, but either different volume levels, or different cards ?



Bluey.

[LonWar]

Just curious.... What bit rate was the mp3 set at?

[crowfan]

Wow, I can't *wait* to see some of the responses to this thread.  

I did the same thing once, and I did hear a difference between the 128 mp3 and the APE. The APE sounded fuller and definitely had a meatier low end. But to me, it wasn't enough to justify the difference in storage space.

crow

[TimB]

Wow, I can't *wait* to see some of the responses to this thread.  

LOL!

-=Tim=-

[Rizlaw]

Bluey,

As 'imjustagamer' noted, it would be nice to know what bitrate you encoded your mp3 test tracks.

I am not familiar with the Chemical Brothers, but this much I can personally say:

1. MP3's recorded at anything less than "alt preset extreme" or 320CBR have a known and measurable drop off in the high frequencies which should be noticable to anyone with good hearing and accustomed to critically listening to non-amplified music at a live concert or well recorded music through a good stereo system (not loud rock because it's too compressed and generally not very well recorded).

2. Sad to say that many young folks and baby boomers have high frequency hearing loss from attending too many loud rock concerts or listening to too much very loud music on their earphones and stereos. It doesn't take long to permanently impair your hearing by listening to loud music on a constant basis . Anyone who hasn't had their hearing professional checked by an audiologist recently, may not realize that they have hearing loss which can affect listening perceptions and judgment.

3. About two years ago, I personally set up a blind demo for the owners of a very high end audio/video emporium in the New York area who are close friends. The reason for the demo isn't important.  I used a few audiophile quality (read - very well recorded material) classical, jazz and pop/rock tracks. Sorry, forgot the albums and artists. I used a then current version of "Exact Audio Copy" in secure mode to rip and encode each track into different formats plus original WAV files with then current versions of:

- Lame mp3 @ alt preset insane (320 cbr);
- Lame mp3 @ alt preset extreme (vbr up to 320);
- APE (lossless); and
- original WAV.

The playback chain consisted of:

- one "Card Deluxe" by Digital Audio Labs - a PCI 24/96 sound card , outputing both analog stereo and coax digital out into a Mark Levinson ML380s preamp > ML332 stereo amp > and Thiel 7.2 speakers.

All tracks were mixed on my computer's hard disc  and played back, at the time, through Winamp (my pre MC days)and only I knew the encoded order of the tracks.

My best recollection of their opinions was that all but 1 agreed that after WAV, lossless APE sounded the best. Lossey mp3 came in last. As to lossy mp3 some preferred CBR over VBR on certain pop/rock selections, but CBR carried the day on all classical cuts.

I don't think all audiophiles are snobs or evangilists for lossless compression; I'm certainly not. Each form of compression has a place. However, while you may or may not agree, lossless encoding (i.e., APE, FLAC, WMA) is quantifiably and audibly better than lossey mp3. It's not a snob or elitist thing, it's a scientific fact thing. Hearing it, can, of course, be another matter, altogether.

You might want to read this article at ExtremeTech about an audio codec shootout they recently had: http://www.extremetech.com/article2/0,1558,1561918,00.asp

Happy listening with MC10.








 

 

[pipsqueak]

Rizlaw

were people able to tell the difference between wav and ape? thats impressive.

pip

[Bluey]

well of course you will hear the difference on a 128 k MP3, or would you ? If you have stuff from the 60s or "lo fi" or something that is a bad recording (not to say all 60s stuff is recorded bad) then MP3 would cover it at these bitrates I think.

I used a preset that generated VBR between 80 to 320.  Usually it drops down to 80 if there is not really much for it to compress, but it sounded ok.

People are right about bit rates though, and I'm tending to stick to higher bitrates, because well I dont care, storage space is not really an issue anymore, and people still try and make bold claims about 64k cd quality hahahaha and go for the ultimate, but well we have 250 gig HD nowdays.  Which is approx 750 CDs encoded in APE assuming 3 CDs per gig!

The only real issue is portable devices, so its a pain to dick about with different bitrates.

But yeah, I still cant tell much difference that would make me whince, I usually go for the slower is better encoding and VBR that tops out at 320k.  I'm not trying to make a comparison with 128k MP3s here.

But i'm not really bothered with ogg or Ape just yet, I have too many MP3s and too many Hardware devices to start the ripping process over again from my 1200 original CDs (ive only done about half anyway).

Bluey.

[Bluey]

Oh yeah dont forget the MP3 encoder you used 2 or even 1 year ago is different to the ones now.  There are still plenty of lameassed MP3 encoders, and people putting stuff out that are encoded lamely (not talking about LAME).

I've done listening tests with people in my house, and I tricked them, making them choose the CD track as the crap one because it was maybe a few DB lower.  Its really important to have the right gear to determine that multiple sources are running through the exact volume.

Yeah but so what, obviously at some point compressed or otherwise audio is going to suffer from lossless, yes it does depend on the material.

I am currently making music with VSTis, I have noticed a difference on my minimoog 5 from arturia running at 44k and 96k, so much so that I'll render my music in 96 k, so i'm looking for a suitable format for songs at 24bits 96k.  However on other VSTis it made no difference, so again its a matter of the material at hand.

Cheers,
Bluey.

[Matt]

Many moons ago I spent lots of time researching what mp3 encoder, version, and bitrate to use.  I also spent lots of time doing A / B comparisons.

I then spent a lot of time ripping my entire collection to MP3.

I was happy for a while.

But improved versions of encoders, new recommend settings, and even new formats kept showing up, making my previous ripping choices less ideal.

Then I ugraded my stereo and found out that what used to sound good didn't sound so good anymore.  That compounded with the fact that with time, it had gotten easier for me to detect lossy artifacts.

So, I switched to lossless.  At the time, it was expensive and there weren't even any playable formats.

It's been a few years.  Now it's less than a dime to store a CD perfectly, and the files are easy to use and well supported.

In my opinion, the angst associated with lossy formats just doesn't make sense anymore.  However, that's just one anal-retentive guy's experiences...

[Rizlaw]

Rizlaw

were people able to tell the difference between wav and ape? thats impressive.

pip

Pip,

As I recall, all 4 individuals seemed to prefer the original WAV (CDA) playback files. We're talking "Golden Eared" music lovers/audiophiles here - these guy listen to showroom $100k+ systems all day long from the top high end audio/video manufacturers. They regularly attend live concerts at the Metropolitan Opera, Carniege Hall and Lincoln Center. They fully appreciate the differences between live and record music.

I think any "conversion process" that changes the original WAV/CDA file, whether it's lossless or not, introduces some artifacts. These artifacts can be additive or substractive and they can be heard by critical ears even though test equipment sometimes can't easily demonstrate the reason critical ears hear a difference. This is the difference between the "Subjective" school of audio listening analysis and the "Measurements Are Everything" school of audio listening analysis which says if you can't measure it, you can't possibly be hearing it.

One only has to read about the ongoing DVD-A (multibit) vs SACD (1-bit ultra high sampling) debate. Both formats are far higher definition audio systems than CD (certainly MP3, OGG, AAC, WMV, etc.) and yet High End Audio Journals seem to prefer the sound of SACD, claiming it sounds more analog, hence, real than DVD-A.

[Matt]

I think any "conversion process" that changes the original WAV/CDA file, whether it's lossless or not, introduces some artifacts. These artifacts can be additive or substractive and they can be heard by critical ears even though test equipment sometimes can't easily demonstrate the reason critical ears hear a difference.

Lossless by definition means there are no changes.  The output from an APE file is bit-for-bit identical to the output from the CD or WAV file.

[EpF]

For me it's a confidence thing - you KNOW when you rip to lossless that you're not going to have any problems in the future, and if/when a lossless codec arrives with better compression ratios, then you can simply convert from one lossless to another.  

I've read that 128kbps CBR mp3s are supposed to be transparent, but then it depends on the genre of music being encoded.  Personally I have plenty of 128kbps mp3s which have noticeable phase-like distortion at higher frequencies.  Apart from that, without perfect hearing, I can still hear a difference in the proportion of frequency ranges on some mp3s.  I have the storage space to keep my cd collection in ape format, and that's future-proof (assuming the hardware doesn't mess up!).  

Strictly in terms of the thread title, there IS a difference, even if you can't always hear it, and what you choose to encode with depends on available storage space and the effect that difference is going to have on what you do with your files...

[MarSies]

Rizlaw,

Me as an ex-audiophile I agree with you completely. I never had the money to by myself a Mark Levinson amplifier set. If i had that might have been my choice!

I can imagine that there is an difference audible difference with wav or ape files. I recall using some stuff to clean CD's from the audio(phile) company were I bought my audioset. We had 2 identical (and original) CD's. We used the cleaning stuff on CD1 and not on CD2. CD1 was significantly different, and better, after cleaning.

This might sound strange but you could hear it. The same goes for a CD player. It's not only the DA converter which is (very) important. It's also the mechanical part. When using good hardware everything sounds cleaner.

Everyone can hear the difference. Just like with a record player and a CD player. I remember tests that i did with hearing the difference between both. Most people preferred the record player. CD's sounded to mechanical and digital. Maybe that has changed with SACD.

I don't understand that everyone wants to go for 128k MP3 or even lower. There is ape and a perfect way to backup your music collection. How much does it cost per CD, hardly anything. For me it's just ape I use. I only use 320kB MP3 on my iHP-140. As soon as the plug-in from SteveG supports ogg I will use ogg with the highest bitrate for the best songs and a lower bitrate for the "normal"songs. Just a matter of space on the device.  


MarSies .....

[Matt]

I can imagine that there is an difference audible difference with wav or ape files.

There can't be.  They're mathematically identical.

If there is a difference, it means something else is happening -- like you're using different digital-to-analog hardware, or different cables, or something similar.

[pipsqueak]

slightly O/T question, but what about the CD player i use to rip the disc in the first place?

there are no audiophile pc cd drives - so can i $200 drive work as well as a $3000 player???

im sold on lossless - but this is my next query...


pip

[gpvillamil]

Here's a funny thing that happened to me. I burned some audio CDs for a friend using MC, starting from LAME VBR MP3s. We popped them into a DVD player connected to the same system as the audio PC to test them, and they sounded BETTER than the MP3s played straight from the PC.

Obvious explanation - the audio DACs in the DVD player are better than the ones in the computer (onboard Realtek AC97).  Unless MC's audio CD burning process has some kind of magic...

[xen-uno]

For those that still think there's a difference...Bluey's "I've done listening tests with people in my house, and I tricked them, making them choose the CD track as the crap one because it was maybe a few DB lower" speaks volumes.

MarSies > We used the cleaning stuff on CD1 and not on CD2. CD1 was significantly different, and better, after cleaning

I don't doubt this if the CD's were in bad shape and thus was forcing the CD player to do a lot of error correction (which involves outright guessing and filling if samples are missing)...otherwise Matt's "something else is happening" applies.

Per the topic though, with a good encoder and a high enough quality setting...there may not be a detectable difference (as in audible) between ape/flac/CD Audio and the lossies. We've come a long way baby.

10^-27

[Rizlaw]

I think any "conversion process" that changes the original WAV/CDA file, whether it's lossless or not, introduces some artifacts. These artifacts can be additive or substractive and they can be heard by critical ears even though test equipment sometimes can't easily demonstrate the reason critical ears hear a difference.

Lossless by definition means there are no changes.  The output from an APE file is bit-for-bit identical to the output from the CD or WAV file.

Matt,

First let me say that I love APE and I use it exclusively for all the CD's I rip to my hard drives with MC10. I consider APE the best audio compressor available and Lame the best audio encoder.  For those interested, http://mp3.radified.com/lossless.htm has a nice website on this subject.

I'm not an electrical/audio engineer, but sterile definitions aside, the point I was trying to make is this: neither APE, LAME nor any other compressor or encoder you care to mention is,  in my listening experience, truly "transparent" when critically compared to the original source CD by a pair of well trained ears utilizing equipment of excellent resolving power. I don't believe that mathematical equivalency between CDA and APE files necessarily goes hand in hand with audible transparency (the copy being totally indistinguishable from the original to the human ear). I am mostly in the "Subjective" camp of audio evaluation. Audio statistics don't always correlate well with what we hear. That said, I agree that APE comes closer to the ideal than anything else I have heard to date.







 

[Charlemagne 8]

Matt,
You should try APE. I think you'll like it.

Everybody Else (you know who you are),

Hearing is subjective, assuming your hearing is still good. You can run great gobs of tests and can probably prove that one type of encoding is better than another. If you can't tell the difference by hearing, that proof is irrelevant. It's an individual thing. Sometimes I think that if you believe that the encoding you are using is superior, that enhances the listening experience.

I'm not running down lossless ... I use APE myself ... it's just that the scientific facts sometimes have no bearing on the perceived reality and the perceived reality is what you're dealing with when listening to music.
If MP3 is just as good as APE to you, use MP3 and save the disc space ... $60 is still a lot of money.

And lossless encoding would be frivolous if you were recording from tape or vinyl. The exception to that would be if you had a 180 gram record, playing on a reference turntable with one of those $3,000 rosewood cartridges. Given that 2 of those records would cost $60 and the equipment would be 100+ times more expensive than that, your hard drive space used would be insignificant.

Merely my opinion ... everybody has one,
CVIII

[JimH]

Merely my opinion ... everybody has one,

Ooh yeah....

[Matt]

I'm not an electrical/audio engineer, but sterile definitions aside, the point I was trying to make is this: neither APE, LAME nor any other compressor or encoder you care to mention is,  in my listening experience, truly "transparent" when critically compared to the original source CD by a pair of well trained ears utilizing equipment of excellent resolving power. I don't believe that mathematical equivalency between CDA and APE files necessarily goes hand in hand with audible transparency (the copy being totally indistinguishable from the original to the human ear). I am mostly in the "Subjective" camp of audio evaluation.

Sorry if I'm being pedantic, but in the digital realm, there is no room for "subjective" analysis.  An audio CD is a string of digital numbers.  A WAV file or APE files is a string of those exact same digital numbers.

Therefore, if there is some audible difference, it is in how you're turning those digital numbers into what you hear (analog).  D/A, oversampling, cables, amp, etc. could all explain those differences.

However, I do agree that lots of things combine to make good or bad sound.  In fact, things like speakers, amps, DSP, cables, etc. may well be more important than the encoder used in most cases.

Matt,
You should try APE. I think you'll like it.

Hehe.  It's hard to trust something with such a stupid name though

[pipsqueak]

yeah yeah yeah ....

but what about the CD player i use to rip the disc in the first place?

there are no ultra-luxury pc cd drives. will my $200 one rip it so it can be played back at the same quality as if the bits were coming out of a $3000 one (assume same amp etc.)?

if the answer is yes, then why would anyone ever buy an expensive CD player?

[xen-uno]

Riz,

Looks like nothing is going to change your mind...facts or otherwise . What exists on a pressed CD is PCM. If you use digital playback in MC with an audio CD...what the soundcard is receiving is EXACTLY the same (bit for bit) as what MC sends when it is decoding any lossless format. By your "utilizing equipment of excellent resolving power", your throwing hardware variables in...and I contend that when making the comparison your adding in more variables still...simply because you need something to decode the apes.

You should level the playing field by equalizing the length of the signal path (source to speakers)...but you would first have to acknowledge that a wav file is an exact duplicate of the data on the CD...do you? You should because it can be proven. If so, then you can ABX between the wav and the ape using your high rez equipment. If you can pick between the two formats consistently, then you substantiate your claims.

Further...assuming you agree that a wav = CD Audio, then your stuck in a quandry...because Matt's point of mathmatical equivalency means that ape>wav equals (exactly) CD Audio>wav.

10^-27

edit: pip...sounds like you bought into that silver is better than copper marketing gimmick. There's alot of reasons a $3k CD player costs $3k...and none of them have much of anything to do with sound quality. A short list...exotic materials/over-engineering/small production runs/marketing/prestige.

[Matt]

yeah yeah yeah ....

but what about the CD player i use to rip the disc in the first place?

there are no ultra-luxury pc cd drives. will my $200 one rip it so it can be played back at the same quality as if the bits were coming out of a $3000 one (assume same amp etc.)?

if the answer is yes, then why would anyone ever buy an expensive CD player?

Just rip in secure mode and you'll be safe.  MC has the best secure mode around (especially with caching drives), so it's what I'd recommend.

[pipsqueak]

thanks matt - id never seen that option before...so whats secure mode?

does it keep re-reading any scratched areas until it gets it right?

pip

[LonWar]

Yes, If it detects something less then 100% it will reread it until it gets it right... Some discs however will never be 100%

[pipsqueak]

thanks gamer

if anyone else out there is like me (bad luck!) and doesnt know much about this, i found this thread useful...


Sauzee test EAC vs MC Ripping in Secure Mode
http://yabb.jriver.com/interact/index.php?board=3;action=display;threadid=15912;start=msg117413#msg117413

pip

[Charlemagne 8]

Quote:
Matt,
You should try APE. I think you'll like it.
 

Hehe.  It's hard to trust something with such a stupid name though  

Nobody's perfect.

[GHammer]

I can imagine that there is an difference audible difference with wav or ape files.

There can't be.  They're mathematically identical.

If there is a difference, it means something else is happening -- like you're using different digital-to-analog hardware, or different cables, or something similar.

I agree Matt. If there is a listenable difference, it would have to be in the decoder causing a delay or some unwanted effect.

But for those with the desire you can prove it to yourself. Use one of the apps that will show you the spectrum of an audio file. Look at the wav and the ape output. Here they are the same with one I tried awhile back.

[Rizlaw]

Riz,

Looks like nothing is going to change your mind...facts or otherwise . What exists on a pressed CD is PCM. If you use digital playback in MC with an audio CD...what the soundcard is receiving is EXACTLY the same (bit for bit) as what MC sends when it is decoding any lossless format. By your "utilizing equipment of excellent resolving power", your throwing hardware variables in...and I contend that when making the comparison your adding in more variables still...simply because you need something to decode the apes.

You should level the playing field by equalizing the length of the signal path (source to speakers)...but you would first have to acknowledge that a wav file is an exact duplicate of the data on the CD...do you? You should because it can be proven. If so, then you can ABX between the wav and the ape using your high rez equipment. If you can pick between the two formats consistently, then you substantiate your claims.

Further...assuming you agree that a wav = CD Audio, then your stuck in a quandry...because Matt's point of mathmatical equivalency means that ape>wav equals (exactly) CD Audio>wav.

10^-27

Xen,

You take me to task about introducing variables, but, in the end, variables are what it's all about in most things. Variables can be good and they can be bad. Each of us makes a choice about the variables we can live with.

Yes, you are correct that it is unlikely that I will change my mind, but not because I refuse to accept the facts as you and Matt may present them, but because I see and hear the "facts" differently. This doesn't necessarily mean that one of us must be wrong. Two + Two = 4, but so does 1+1+1+1.

I agree that a WAV file can = a CDA file assuming the bits are copied without error to the hard disc.  However, as I understand it, CDrom drives do not have the same robust Reed-Solomon error correcting capabilities as standalone cd transports/players have to correct digital soft errors, hence error correction appears to be more of an issue with PC CDrom drives. PC DRAM adds another layer of variability, as does the hard disc with write errors (I realize this may all seem to be nit picking - but numbers don't always reveal the truth about things - I'm told you learn that in Statistics). That's why, when it comes to audio issues, I'm not always willing to accept arguments based solely on numbers when my ears and experience tell me otherwise. Build a stereo component strictly by the numbers (without critical listening tests) and you can almost guarantee a horrible sounding component. Bottom line: I'm not a fan or believer in the 'bits is bits' school of thinking because it assumes everything is "perfect" and that's not the real world.

I also don't necessarily accept that an APE compressed file = CDA uncompressed file, and hence sounds 100% transparent. If there is a perfect compressor out there, I have yet to hear any authoritative group proclaim its existence. I sure would like it to be APE though.

As for my own testing. As I have said, I love APE. I use APE. I think it's as good as we can get for now, but it's not audibly 100% transparent 100% of the time to my ears.  When I make these personal assessments, I set MC10 to output a stereo digital signal to my Card Deluxe sound card which has had it's Goldmund coax and Accuphase optical digital outputs connected, at various times, to Mark Levison equipment, Proceed equipment and a top of the line Denon AVR5803 receiver. (I did mention I'm a recovering audiophile didn't I?).

I can and do hear subtle differences between WAV and APE output from the computer to my Thiel 7.2 speakers; albeit not 100% of the time. Note that I said subtle. I'm sure there are enough variable in my computer to account for some of these audible differences.  When playing APE files from the computer though my stereo system vs. playing the pressed CD on my standalone equipment, the differences are more noticable every time.  Since we can't burn and play native APE files on standalone players, there is no way to tell if APE would sound the same as a pressed CD on a stereo system outside of the computer environment.

The bottom line here is that audio is first and foremost an  emotional listening experience, if you like what you hear stick with it, if not, change, but you shouldn't let numbers be the sole arbiter in your decision making process; they can, and often do, lead you astray.  

   

[gpvillamil]

yeah yeah yeah ....

but what about the CD player i use to rip the disc in the first place?

there are no ultra-luxury pc cd drives. will my $200 one rip it so it can be played back at the same quality as if the bits were coming out of a $3000 one (assume same amp etc.)?

if the answer is yes, then why would anyone ever buy an expensive CD player?

You got it right... Meridian Audio, for example, uses a PC drive in their high-end players. They read into a large RAM buffer and play out from there. The buffer takes care of any issues with jitter, latency, etc. Now, the quality of their DACs is very, very good so there is a premium there.

But as for putting lots of money into the mechanical transport? Can't see the reason. Cheap PC drive into a buffer can perform as well.

[TimB]

Yes, If it detects something less then 100% it will reread it until it gets it right... Some discs however will never be 100%

Right.  It reports what percentage it did achieve so you can see what goes 100 and what doesn't.

-=Tim=-

[ZenDav]

  My two cents...

  Wav files can reliably, predictibly and consistantly translate the same exact digital waveform samples to and from Ape files (and flac).  If you can hear differences between matching wavs and apes on a given PC with the same music program, sound card and speakers, it indicates an error in the music program which should be sending the exact same data samples to the sound card driver.  (or you transcoded in error?)

  As digital sound captured on CD is done with barely adequate digital error detection and correction data (data CDs dedicate ~150MB more per CD), quality retrieval is dependent on quality read optics and associated error detection/correction circuity.  This can vary greatly, but far more varible is the waveform created internally by the CD drive itself (non data mode) and/or by attached sound circuits and the subsequent path to speakers.  An audio comparison between of any lossless format of audio file played in MC on a pc and a CD played in high end audio equipment would be unfairly leaping to conclusions to determine it is the file format that was somehow subtly different.

  Whereas MP3 files are quality attempts to produce very "sound alike" waveforms in data sizes far smaller than were used to store the raw sample data, APE files are the exact same data samples stored in less data via complex data reduction algorithms (both wrapped with format and attribute data).  MP3 file size is based on the varible slide of quality verses size with higher bit rates being able to represent a more accurate "sounding" waveform.

  So if the original question was do 128bps MP3 files represent the same waveform as APE files? No.  If it was simply can anyone audibly distinguish between high quality renditions of the two? Yes.

  How about the quandry, does the observer changes the object to be observed by his observation.  Or maybe it's just a sollipsism.  

[Alex B]

I can sometimes hear the APE/MP3 difference. One recent example was when I ripped a classical symphony and encoded both formats: APE and LAME --alt preset extreme. At first they sounded identical. After a while I started to notice subtle occasional differences. Some concert hall echoes, quiet instruments and crowd noises in the background at the same time when the orchestra was playing loudly sounded a bit different. Those weak sounds were more natural in APE. I must point out that the differences were minor and very careful listening was needed for finding them.

There can be much bigger differences between different hardware. For example:

I have a five-year-old stand-alone Sony CD player. Some time ago I bought a Terratec DMX 6fire 24/96 sound card. It's obvious that Terratec has better DA converters. My old CDs sound now better after ripping than before when I played them directly with the Sony player. So my MP3s and APEs are better than the originals in this case  .

I have also a stand-alone DVD player. It has even worse DA converters than the Sony player. Standard audio CDs don't quite have the same audio quality and its MP3 playback option is really awful -- probably caused by a bad decoder combined with a bad DAC.

The sad thing is that just a few days ago the digital outputs of my Terratec sound card stopped working and I have to send it out for a warranty repair.

I just uninstalled the Terratec card and enabled the integrated C-Media 9739 device. I had not tried it before. I connected it digitally with my Yamaha receiver and could instantly hear a difference. It sounds somewhat distorted and harsh. The C-Media device resamples all material to 48 kHz and that must be the reason. Maybe I should try its analog output too. I can imagane how bad it would be...

I hope my Terratec gets fixed soon.

[Bluey]

Yes its true, many people compare MP3s with their standalone CD players, and at the same time have crap soundcards in making a comparison, usually the soundblasters or now the popular on board sound chips on the mobos (mine is totally awefull sounding and full of noise).  Fortunetely I have a new EmuX (120 db) card and Pulsar, both basically have mastering performance output.  But I think maybe the Audigy 2 is a bit better but I do not know.  So a lot of people I know based assessments on not having the right DAs and soundcard and also no understanding about DB levels between different hardware devices.

I have a sony CD player which was top of their line, but its 12 years old !  When I put my PC output and CD player through my mixer the PC sounded much better playing the same stuff.

So their you have it, I think many people are now managing to get better soundcards at a cheap price, whilst holding on to old CD players.  I think most peoples CD players are maybe 5+ years old.  But these soundcards on a mobo are truly awefull except for games.

An added bonus is that I can use MC DSP plugins, Izotope Ozone for winamp (which works with MC and is pretty good) or Dee 3 or whatever which certainly make a difference.

Bluey.

[Alex B]

Actually that C-Media sound circuit (in digital mode) is not bad at all when compared to older PC soundcards. There is no noise, all the basses and trebles are present and there is no audible harmonic distortion. Probably it sounds like an older entry-level hifi CD player.

So, in my vocabulary that "huge difference" I mentioned before means "just noticeable difference".  

[Shady Bimmer]

You take me to task about introducing variables, but, in the end, variables are what it's all about in most things. Variables can be good and they can be bad. Each of us makes a choice about the variables we can live with.

Sounds like some cloudiness.  What would some of these "variables" be?  Once you throw unknown external variants into the mix, comparing audible qualities between lossy and lossless compression formats becomes invalid.  Comparison between any two formats is in fact invalid at that point.

Yes, you are correct that it is unlikely that I will change my mind, but not because I refuse to accept the facts as you and Matt may present them, but because I see and hear the "facts" differently. This doesn't necessarily mean that one of us must be wrong. Two + Two = 4, but so does 1+1+1+1.

"see" the facts differently?  Since you brought it up. . .how about something like this:

• Start with 2 channel, 16 bit, 44KHz PCM audio sample (.WAV)
• Use the best MP3 encoder you can find at its highest setting (.MP3)
• Use the best APE encoder you can find (.APE)
• Use an audio spectrum analyzer to compare all three
• Use MC to play all three via digital out from the same sound card in the same PC.  Capture the output and compare the results from all three.

The capture of the PCM stream from the audio cards will be difficult for most home users, but the first comparison will work just as well.

You should find that the APE and WAV graphs are identical and the MP3 graph will differ.  This has been done and this is the case.  If you "hear" a difference between APE and WAV then there is another factor involved that is unrelated to the lossless format.

PC DRAM adds another layer of variability, as does the hard disc with write errors

Hard disk with write errors?  Any non-detected write errors on a hard disc, cdrom (not cd-audio format, which would be a typical music CD), or other permanent storage device is considered a failure and results in corruption.  If your hard disc did not reliably store data your OS (windows, linux, macos, etc) would fail.  Drive circuitry is responsbile for ensuring that what gets written to it is exactly what gets read. . .with an indication when this is not possible.

I also don't necessarily accept that an APE compressed file = CDA uncompressed file, and hence sounds 100% transparent. If there is a perfect compressor out there, I have yet to hear any authoritative group proclaim its existence.

Now I'm totally lost.  What do you consider a "perfect compressor"?  A "lossless" compressor (of which APE is one) takes a source and simply stores it in another format using less space with no loss of information.  Period.  In other words, start with file 'A'.  Compress with APE to create file 'B'.  Decompress 'B' to create file 'C'.  Compare 'A' and 'C'.  By definition 'A' and 'C' will be bit-for-bit identical.  As Matt has already said - if they are not then there is something else wrong.

If you try this using MP3 instead of APE, you will find that 'A' and 'C' are absolutely different.  MP3 is a 'lossy' format and by definition it throws away information to reduce size.

Given that the D/A converters and analog circuitry are common when playing WAV, APE, MP3, then any audible differences between APE and WAV are purely psychological.  The source into the D/A is identical, and everything beyond that point is common.  An MP3 will result in a difference since the source will be different.   The extent of the audible difference is purely subjective when dealing with lossy formats and there have been many works written about this subject.

So, if being able to recreate the original source bit-for-bit identically does not meet your definition of 'perfect compressor' what is your definition?

When playing APE files from the computer though my stereo system vs. playing the pressed CD on my standalone equipment, the differences are more noticable every time.  Since we can't burn and play native APE files on standalone players, there is no way to tell if APE would sound the same as a pressed CD on a stereo system outside of the computer environment.

Ah - but this difference is not due to APE vs WAV - it is due to completely different sets of DACs, cables, and possibly other analog equipment.  It is known that even two same-part DACs from the same manufacturer can produce slightly different analog waveforms from the same digital source (though not likely audible) so unless everything including and between Digital/Analog conversion and the listener's ears is common you absolutely can not make any claims about the digital source sounding different.  This is a fundamental of conducting testing/experiments.  Your testing has far too many variables to make any claims about any single component.

While you may not be able to play APE on your standalone player you certainly can play WAV on your PC.  RIP a CD to your PC.  Encode to APE.  Use MC to play both the ripped WAV and the encoded APE.

Better yet - perform a proper double-blind experiment and see if you can still hear a difference.

[enigman]

Interesting stuff all around.

But for those of us who still have that high-school-science-class mentality and what to try everything out first hand, does anyone know if there is a program out there that will allow me to blind-test different audio formats?  I could try to fake it myself, but I know that unless I really really truly really don't know before hand which file is which, I won't be able to do a truly blind audition.  Is there some nifty piece of software that will let me pick 2 files and then randomly play them back to me while recording which one I like better?  That'd be cool.  Also, does anyone have a link to a discussion on preferred methodology for blind audio testing?

Thanks

--Tim

[Alex B]

a link to a discussion on preferred methodology for blind audio testing?

Some links:

http://doc.hydrogenaudio.org/wikis/hydrogenaudio/ABX ,

http://www.hydrogenaudio.org/index.php?act=ST&f=5&t=7516#entry74066  and

http://www.hydrogenaudio.org/index.php?showtopic=12118

[enigman]

Sweet.  I'll have to check this out when I've got a free moment.

--Tim

[Alex B]

I found two more in my I.E. favorites:

http://www.bostonaudiosociety.org/bas_speaker/abx_testing.htm

Jussi Hynninen: A software-based system for listening tests
This Master’s Thesis has been submitted for official examination for the degree of Master of Science in Espoo on May 31st, 2001

[Rizlaw]

I thought this might be of some interest for those who are of the opinion that lossless compression is perfect:

"Audio files contain a certain amount of information - "entropy" - so they cannot be compressed losslessly to any size smaller than that.  So it is not realistic to expect an ever-increasing improvement in lossless compression algorithm performance.  The performance can only approach more closely whatever the basic entropy of the file is.  No-one quite knows what that entropy is of course . . . I think that would require understanding the datastream in a way which is exactly in tune with it's true nature.  For instance a .jpg image of handwriting would appear to contain a lot of data, unless you could see and recognise the handwriting and record its characters in a suitably compressed format.  The true nature of sound varies with its source, physical environment and recording method, and a lossless compression program cannot adapt itself entirely to the "true" nature of the sound in each piece of music.  Therefore it is not surprising that different algorithms work best on different kinds of music."

The above quote comes from  Robin Whittle's website "First Principles":

http://www.firstpr.com.au/audiocomp/lossless/

The part I found interesting was the sentence:

"The true nature of sound varies with its source, physical environment and recording method, and a lossless compression program cannot adapt itself entirely to the "true" nature of the sound in each piece of music."

I'm sure there will be some who will consider Robin Whittle's opinion to be more metaphysical than scientific.

[Matt]

"The true nature of sound varies with its source, physical environment and recording method, and a lossless compression program cannot adapt itself entirely to the "true" nature of the sound in each piece of music."

I'm sure there will be some who will consider Robin Whittle's opinion to be more metaphysical than scientific.

It's scientific, but don't confuse quality and size.

Robin is saying that because lossless compressors can't perfectly adapt, their compression is sub-optimal with regards to size.

However, they still maintain perfect quality in all cases.

I wrote this a long time ago, but maybe it'll make some sense about how a lossless compressor saves space while staying bit perfect when decoded:
http://www.monkeysaudio.com/theory.html

[gpvillamil]

I thought this might be of some interest for those who are of the opinion that lossless compression is perfect:

"Audio files contain a certain amount of information - "entropy" - so they cannot be compressed losslessly to any size smaller than that.  So it is not realistic to expect an ever-increasing improvement in lossless compression algorithm performance.  The performance can only approach more closely whatever the basic entropy of the file is.  No-one quite knows what that entropy is of course . . . I think that would require understanding the datastream in a way which is exactly in tune with it's true nature.  For instance a .jpg image of handwriting would appear to contain a lot of data, unless you could see and recognise the handwriting and record its characters in a suitably compressed format.  The true nature of sound varies with its source, physical environment and recording method, and a lossless compression program cannot adapt itself entirely to the "true" nature of the sound in each piece of music.  Therefore it is not surprising that different algorithms work best on different kinds of music."

The above quote comes from  Robin Whittle's website "First Principles":

http://www.firstpr.com.au/audiocomp/lossless/

The part I found interesting was the sentence:

"The true nature of sound varies with its source, physical environment and recording method, and a lossless compression program cannot adapt itself entirely to the "true" nature of the sound in each piece of music."

I'm sure there will be some who will consider Robin Whittle's opinion to be more metaphysical than scientific.

Actually I'm familiar with Robin Whittle's work, and his point is very clear and simple.

Basically, he agrees that lossless compression is in fact, lossless. Which for our purposes means perfect. His focus is 100% on achievable compression ratios, not sound quality. After all, if the resulting decompresssed file is bit for bit identical, there is no point in going any further. Note that on his website he spends no time dealing with audio quality issues, as he accepts that these are irrelevant in this context. All the results he presents are regarding the achievable compression ratios.

What his quote is referring to is this:

There is a certain basic amount of information in a file (which he calls "entropy") which sets an absolute lower bound on the ability to losslessly compress it. So he is saying that he does not expect lossless compressors to be able to improve their performance dramatically in terms of THEIR ABILITY TO ACHIEVE SMALLER FILE SIZES.

His comment about an understanding of the nature of the piece refers to two phenomena. The first is that different algorithms perform better (achieve smaller file sizes) on different types of music. This is understandable - after all, that is why we have lossless compressors specifically designed for music, and don't just use ZIP or TAR or RAR. A compressor with some awareness of the nature of the data will usually perform better. [Edit] The most obvious example of this is the correlation between left & right channels of a stereo recording.

The second point is more subtle, and refers to a situation where you use a deep understanding of the nature of the information to achieve very high compression ratios. In his example, a sample of handwriting encoded as a JPG (or as a bitmap) would appear to contain a lot of data for a compressor to work on. However, if you interpreted that same sample of handwriting as a sequence of vectors, the information would be a lot more concise! In the same way, if you try to compress an audio file which is a digital representation of the captured sound, it will contain a lot of detail. However, if you represent a given performance as a score and an arrangement, the amount of information is a lot less! The extreme example of this is "semantic compression", where you rely on a shared understanding of what is being sent to achieve very high compression levels.

So the comment about being able to achieve greater compression through a true understanding of the nature of the piece refers to the latter: if you actually understand the nature of information being conveyed, you can achieve very high compression ratios vs just treating it as a stream of bits.

Now, the amount of information you actually need to send in order to make this kind of compression "lossless" is a matter of debate... In the handwriting example, in addition to the strokes, you would also have to send information about the nature of the pen, the texture of the paper, lighting conditions, etc in order to reproduce exactly the bitmap representation.

But to sum up: there is nothing metaphysical about Robin Whittle's analysis. He agrees that lossless is for all intents & purposes perfect. His focus is on achievable compression ratios. There is an absolute limit on achievable *lossless* compression ratios. Getting around this, or even approaching it closely, requires a good understanding of the nature of the source material.

[DougHamm]

I thought this might be of some interest for those who are of the opinion that lossless compression is perfect:

...

The part I found interesting was the sentence:

"The true nature of sound varies with its source, physical environment and recording method, and a lossless compression program cannot adapt itself entirely to the "true" nature of the sound in each piece of music."

I'm sure there will be some who will consider Robin Whittle's opinion to be more metaphysical than scientific.

Whittle's point is that sound is too variable to make hard-and-fast rules by which to compress it losslessly to any great degree.  His point is not whether it sounds like the original, but at what size must the lossless file be in order to reproduce the original bit-for-bit.  MP3s are a tradeoff of quality vs. size, but with lossless formats the only variable is size.

Again it all comes down to the very simple litmus test of a bit-for-bit comparison of a .wav and an .ape, or .wma (lossless), or .flac, or .shn.  They all come out the same, their only variation being relative size due to the efficiency of the methods they employ to "approach entropy".

-Doug

[fex]

...Robin is saying that because lossless compressors can't perfectly adapt, their compression is sub-optimal with regards to size...

Taken from another thread
(http://yabb.jriver.com/interact/index.php?board=3;action=display;threadid=20257):

A lot more :-)
My APEs tend to range from 500-900 kbps, or 3-5 x the size of 192 kbps MP3s.

So don't forget your harddisks. And the $ to buy them.

[Fixer]

I don't believe that mathematical equivalency between CDA and APE files necessarily goes hand in hand with audible transparency (the copy being totally indistinguishable from the original to the human ear). I am mostly in the "Subjective" camp of audio evaluation. Audio statistics don't always correlate well with what we hear. That said, I agree that APE comes closer to the ideal than anything else I have heard to date.

New guy here....

   If a WAV file converted to APE and then converted back to WAV, and is found to be bit for bit identical to the original file, it will sound *exactly* the same.  There is no other possible outcome of the listening test.  If anyone hears a difference, then there are other factors involved, including possible preconcieved ideas of the listener.

   That's the cool thing about digital, if all the stored numbers are identical, the content IS identical.  Although it may take years, if you read the numbers from the file to a friend around the world, and nobody made any mistakes writing them down, the friends copy across the world would have an *exact* copy, and sound exactly the same (assuming identical equipment, speakers and room acoustics).

   I can *easily* hear the difference between 128 mp3 and ape/wav,  At 256 and higher, I just don't know, meaning it may not be important.  However, what if my ears get better, or I learn to listen better tomorrow, or I get better equipment and the difference is very apparrent, I can't go back, lossy is forever.  I stick to ape to never limit my choices for the future.  And I don't think mp3 can do a totally gapless track switch.

[Rizlaw]

Actually I'm familiar with Robin Whittle's work, and his point is very clear and simple.

Basically, he agrees that lossless compression is in fact, lossless. Which for our purposes means perfect.

But to sum up: there is nothing metaphysical about Robin Whittle's analysis. He agrees that lossless is for all intents & purposes perfect. His focus is on achievable compression ratios. There is an absolute limit on achievable *lossless* compression ratios. Getting around this, or even approaching it closely, requires a good understanding of the nature of the source material.

 gpvillamil,

I found your explanation of R. Whittle's work lucid and quite helpful. Perhaps we can put the issue to bed now.
 

[MarSies]

I missed  one day and this thread has become very long. One of the best threads I think. Don't think that it will be "solved".  

Difference between ape and wav, I don't know which variables are there to make it sound differently. Maybe because there is more conversion to be done which draws current which causes other electronic parts to work slightly less good. This is not just crap. On "better" CD systems you can switch of the display to prevent distortion and other abnormalities. Yes it sounds strange but it works. (also different power supply's)

How many people say that a digital system is better then the old analog way. Yes theoretically that's true. In real life a good record player and record beats a normal CD by far. Loads are CD's are recorded from analog tapes or just from records. Not always with old music.

Don't think wel will ever know all the variables which makes something sounds like it does.

Rizlaw, I would really like to see your system. Haven't been around in the audiophile world for a long time. Would like to do some more but that would involve spending money I don't have.  


MarSies .....

[cryst]

I use APE exclusively.  My files sound exactly like the originals.  I waste no time wondering how much quality I'm losing, comparing this-to-that, or asking other people if my files really do sound like they're supposed to.  I already know they sound like they're supposed to.  Hard drive space is easy to come by these days.  All this makes it a no-brainer decision to me.  It's that simple.