Help - Search - Members - Calendar
Full Version: Flac vs. Apple Lossless from WAV 24Bit 96kHz
Hydrogenaudio Forums > Lossless Audio Compression > Lossless / Other Codecs
bytemission
Hi,
first i have to say thank you for this great forum!

I have tried some things regarding to compress lossless my complete collection of digitized records.
I usual record them in 24Bit/96kHz WAV.

So i tried to compress that to Apple Lossless and Flac using different sopftware. All on Apple macintosh: iTunes for Apple Lossless and Max 0.7 for FLAC5 + 8 and Apple Lossless as well.
After compressing i decode again back to WAV and the i checked with the command lind diff command if the files differ: ALL DIFFER, THEY DIFFER FROM ORIGINAL AND FROM ALL OTHER FILES. Hm, really bad or better to say: not what i have expected...
Then i tried to recode wav to wav the same way i decoded the files: same result.

The md5 hashes give nearly the same result with one exception: encoded to Apple Lossless with Max 0.7, decoded with Quicktime (for decoding i used Quicktime, because this seems to be the only software where i get back 24Bit! /96kHz) and compared to the recoded wav from original wav have exactly the same md5 hash. Soe these 2 are really really the same.

So is "diff" the wrong software to check that out?

Please help me... Don't want to loose quality by converting....

Thank you!

knutinh
Could you inspect the files (prefereably the sample-by-sample difference) in a waveform editor to see if one program is adding a few samples of silence at the start, doing normalising or some other strange thing?

-k
bytemission
QUOTE(knutinh @ Mar 15 2007, 13:36) *

Could you inspect the files (prefereably the sample-by-sample difference) in a waveform editor to see if one program is adding a few samples of silence at the start, doing normalising or some other strange thing?

-k

hi, yes i think i can. but can you explain me what to do exactly?
thnx
2Bdecided
The most likely explanation is that the headers have changed, while the audio data is identical.

Especially for greater than CD audio quality .wav files, there are two different but correct header formats.

Cheers,
David.
pepoluan
QUOTE(bytemission @ Mar 15 2007, 20:40) *
QUOTE(knutinh @ Mar 15 2007, 13:36) *

Could you inspect the files (prefereably the sample-by-sample difference) in a waveform editor to see if one program is adding a few samples of silence at the start, doing normalising or some other strange thing?

hi, yes i think i can. but can you explain me what to do exactly?
If you're on Windows, use foobar's bitwise comparison.

If you're not, grab an audio editor, do a paste-mix-inverted of one sample onto another sample. The resulting waveform should be a flatline 0.

bytemission


If you're not, grab an audio editor, do a paste-mix-inverted of one sample onto another sample. The resulting waveform should be a flatline 0.


thanks. in all cases it's flat! so all is okay wink.gif thank you for your great help!

best wishes

bytemission
ffooky

Much easier than fannying around with editing software would be to use shntool's cmp function, either from the command line or via the shntool tab of xACT.
MedO
QUOTE(bytemission @ Mar 15 2007, 13:14) *

The md5 hashes give nearly the same result


Just want to add, md5 hashes are either the same or they are not. If you have two similar (but not matching) hashes, that gives you no information about the similarity of the original files, just that they are different. In other words, even if you change just one byte in a file, the new md5 hash will (probably) be completely different from the old one.
pepoluan
QUOTE(ffooky @ Mar 17 2007, 07:40) *
Much easier than fannying around with editing software would be to use shntool's cmp function, either from the command line or via the shntool tab of xACT.
Rats huh.gif I *really* should be studying the shntool documentation wink.gif thx for the heads-up.

frankiecc
QUOTE(bytemission @ Mar 15 2007, 07:14) *

Please help me... Don't want to loose quality by converting....


I don't know how you tested this but my tests show that Apple Lossless is what it says it is: lossless.

MD5 fingerprints (checksums):

Song (Original).wav ---------------------> 4f2098e2aa5db7b52810500f6dd07bba

Song (Original to Lossless).m4a ---------> c551eab61294558f1ba043219cd951a5

Song (Orig. to Lossless to Orig.).wav ---> 4f2098e2aa5db7b52810500f6dd07bba
bytemission
did you test it with 24bit/96khz?

btw. after having all in apple lossless now i found that there is no application at the moment what can do a batchconvert back to wav/aiff without loosing the bitrate (all apps converts down to 16bit or crash, f***) at the moment.
apple soundtrack: no
amadeus pro: no
itunes: no
max: no

the only app for convert to wav / aiff without changing the bitrate is: quicktime itself sad.gif

i know, it's my fault, i had to check that first. but as i saw i can make applelossless i did not expect to endup in a one way road...

if anybody has an idea please tell me wink.gif thnx

MD5 fingerprints (checksums):

Song (Original).wav ---------------------> 4f2098e2aa5db7b52810500f6dd07bba

Song (Original to Lossless).m4a ---------> c551eab61294558f1ba043219cd951a5

Song (Orig. to Lossless to Orig.).wav ---> 4f2098e2aa5db7b52810500f6dd07bba

[/quote]
sbooth
QUOTE(bytemission @ Apr 25 2007, 13:38) *
did you test it with 24bit/96khz?

btw. after having all in apple lossless now i found that there is no application at the moment what can do a batchconvert back to wav/aiff without loosing the bitrate (all apps converts down to 16bit or crash, f***) at the moment.
apple soundtrack: no
amadeus pro: no
itunes: no
max: no

This is not correct. Max can convert Apple Lossless at any bit depth to AIFF at any bit depth. However, depending on how the AIFF output is set up, you can get dithering (ie 24 -> 16 bit or even 8 bit). Be sure you've selected 24-bit big endian signed integer as the AIFF format type.

What do you mean by losing bitrate? Since the process is lossless, there can be no change in bitrate. Whether applications are displaying the bitrate correctly is a different animal, though. iTunes historically has a problem displaying bitrates for MPEG-4 AAC files created with Max, but I believe this is fixed in the latest iTunes.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.