WAV or Apple Lossless Encoder?

Peter_s and Steve - I would like to understand just what extra information is in an ALAC file that degrades its sound. Any explanation or references to this? I know that this is a proprietary format, but if the claim is that there is some extra data in it, it would be great to have some idea or what that is and how it is used. References?

Jax2 - I am not a MAC user, sorry. But EAC and foobar both have comparison routines in them for wave files. Do I read your post correctly that you are using a MAC to stream to a Transporter? The Transporter is yet another wrinkle in the comparison - not sure exactly what it uses for ALAC decoding.

I really think that concluding that iTunes is inferior for ripping files (at least from relatively clean disk) based on these observations is premature. I, and others, have ripped files with iTunes, converted them to wave and they are EXACTLY identical to wave files produced from EAC - every bit is the same. And a 50 MB wave file has a lot of bits! Two wave files with exactly the same bits cannot sound different when played by the same player. If somehow they do sound different, then something else has happened in the process. I just do not see how iTunes ripping can be inferior when it produces exactly the the WAVE file as EAC? Incidently, I do believe that EAC can do a better job on a damaged disk - but of course none of us has those :)

If a given player playing an ALAC file and playing a WAVE file produce different sounds, it is a big leap to believe that the only cause of this could be the ripping process. It may well be how the player somehow handles the different formats.

My background is scientific research and in computer science. So, when I hear that there are differences in 2 approaches I want to know why. And I tend to get picky about experimental design. In looking at the results that people are reporting I see lots of unaccounted for variables. The experiments need to be designed to eliminate those variables. My suggestions, although it will take time, is to

1) Compare the sound of ALAC files ripped with iTunes (using the error correction mode) and ALAC files that were ripped to WAVE by EAC and converted to ALAC by iTunes - both played with iTunes (or your favorate player). If you are using something other than iTunes, you should probably use the same conversion software as your player uses. It is hard to do bit compares on these ALAC files, unfortunately. Waves are much easier.

2) Compare the sound of WAVE files ripped with iTunes (directly to WAVE) and WAVE files ripped with EAC. This should probably be done with iTunes as well as with some other player like foobar or J River. Foobar is a good choice since it has a ABX comparison function to allow for blind testing. When I have done the ABX comparisons with foobar with EAC and iTunes produced wave files, I cannot reliably tell the differences - although I admit I have not done it a lot of times and blinding test is an art unto itself. If you have not done it, give it a try. It is a interesting experiment.

3) Repeat 2) with files ripped with iTunes into ALAC and then converted to WAVE by iTunes (rather than ripped to WAVE directly).

Its important to compare apples and oranges and to be sure to eliminate outside influences on what is being tested - in this case ripping accuracy.

Now off to listen to some music.

I experimented with two tracks, one from Diana Krall and one from Keb' Mo'.

I already had these on my iPod in Apple Lossless, I use a Wadia iTransport.

I copied these two tracks using EAC as .wav files and copied them into iTunes. I then used the iTunes tool to convert them to Apple Lossless, but keeping the .wav version also, giving me 3 copies to of each; I also copied the Diana Krall in for a 4th version, this using iTunes creating a .wav version.

In playing these I was thinking there was a slight difference, with the original iTunes Apple Lossless version being very slightly of lesser quality, although I played the cuts many times questioning this. This surprised and disappointed me. Then I realized the "original" version I created was done back before I started using error correction and wondered if that could be the cause of any difference. So I replaced the original with a newer Apple Lossless version with error correction on. My conclusion is that I can not hear any difference.

Maybe my song choices weren't the best, maybe you ears are better or maybe _____________, you fill in the blank. If I get a chance this weekend maybe I'll give yet another listen. When doing these comparisons it becomes quite annoying.

Brian

Here's the info that SqueezeCenter provides on the three files Peter provided to me last night:

EAC to AL conversion: Bitrate 634kbps VBR / Volume Adjust -6.97db / File Length 11,640,516

My WAV File: Bitrate 11411kbps CBR / File Length 25,279,340

Peter's AL file: Bitrate 634kbps VBR / Volume Adjust -6.97db / File Length 11,639,656

I don't think I have any software that compares file size, but Peter has PC's as well as Macs and may be able to compare bit-for-bit info.

I listened again this evening and there's no doubt of the differences I heard last night. I also threw in the cut played via digital input from a CD transport to the same Modwright Transporter DAC. The EAC copy sounded most like the transport and I'd have a very difficult time telling the two apart. I could certainly tell the difference between the WAV and AL files, from the file playing off the disc.

Just for an experiment I went downstairs to my office system, where my music library is stored. I played the same cuts on a much less revealing system, via iTunes. There I could not tell the difference between the three files, and like Brian, I found the whole thing frustrating after a short while.

I don't think the difference is worth ripping your entire library over again. It most certainly is not worth it if you have to go in and label all the cuts on every CD manually, as it seems would be the case ripping to EAC and converting to AL as Peter described.

Dtc - yes, streaming files via Wifi to Modwright Transporter. No idea how it handles ALAC conversion. It sounds damn good though.

Interesting subject. I too thought bits are bits and that lossless meant just that.

Dtc - a few responses to your most recent post:

"I would like to understand just what extra information is in an ALAC file that degrades its sound. Any explanation or references to this?"

Steve will have to answer this. I was merely repeating what he told me.

If a given player playing an ALAC file and playing a WAVE file produce different sounds, it is a big leap to believe that the only cause of this could be the ripping process. It may well be how the player somehow handles the different formats.

Note that both Marco and I are independently hearing a difference in sound between two files of the same format (ALAC) that were generated differently. This cannot be how the player (itunes) handles the format. BTW - I am understanding that itunes does the conversion from ALAC to uncompressed, and sends the 16/44.1 stream to my Pacecar.

My background is scientific research and in computer science. So, when I hear that there are differences in 2 approaches I want to know why. And I tend to get picky about experimental design. In looking at the results that people are reporting I see lots of unaccounted for variables. The experiments need to be designed to eliminate those variables.

I am so with you on that. I too am in a science/engineering field.

My suggestions, although it will take time, is to
1) Compare the sound of ALAC files ripped with iTunes (using the error correction mode) and ALAC files that were ripped to WAVE by EAC and converted to ALAC by iTunes - both played with iTunes (or your favorate player).

This is exactly what I did that started my entry into this thread after hearing the difference. Now - I'd like to ask you to do the same thing! I know you've made bit/bit comparisons of the WAV's - but try doing this and listening and tell us what you think. To eliminate another variable - try the song Marco referenced above. Or, if you have an FTP site, I can send you both versions. Would you be willing?

Unfortunately, to do your second method I'd need my CD, which I accidentally left at Marco's the other night. It will take me a while to get it back. But please, let us know what you find out if you make comparison #1 above (as Marco noted, you must have a sufficiently resolving system).

Finally - personally, I would try to re-rip music as I listen to it, not as a big project. BUT - I don't yet know what will be the right ripping procedure to accomplish this. Part 2 of this exploration!!!

Look, I am a EE with over 30 yeard design experience. We can debate the theory and the science until the cows come home. The fact is there are differences here, and comparing .wav to .wav files does not explain it.

One person that did this experiment repeated it with a Transport rather than the Pace-Car reclocker and concluded that if the jitter is very high, you really can't tell the difference. It's when the jitter gets really low that the difference is obvious. The Pace-Car is only a FIFO, it does not change the data in any way.

Steve N.
Empirical Audio