Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Lossless vs. Redbook tests? (Read 116542 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Lossless vs. Redbook tests?

I was referred to this site by a friend in another computer hi-fi site because that one is starting to get knee deep in the all of the subjective, touchy-feely audiophile stuff.  I've searched this forum as best I could but cannot not find any test results for comparing the sound quality of lossless compression (FLAC, ALAC, etc) files to the uncompressed file (Redbook, WAV, AIFF).

I did find a couple of threads on the subject, and the answers were usually a terse "lossless = no loss."  Which is fine and dandy, but I'd like to see the actual tests that lead these people to this conclusion.

My motivation is this comment made in that other forum:

"I know of some very knowledgeable engineers who have measured differences between Apple Lossless files and AIFF files. They measured the files, once a sonic difference was heard in an A/B comparison, in the first attempt to quantify the difference. The measurements were carried out on several computers with some of the best test equipment available."

Thanks much,

- Tim

Lossless vs. Redbook tests?

Reply #1 –
Yawn...

Lossless vs. Redbook tests?

Reply #2 –
I would suggest that you look up the definition of the word "lossless".

It really is that simple.

[edit] or, alternatively - encode a WAV / AIFF file in your lossless codec of choice, decode to a new WAV / AIFF file and compare the audio with the original.

Again, it really is that simple.[/edit]
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

Lossless vs. Redbook tests?

Reply #3 –
You could have found a million threads on this site with one search on this subject.

I.e. 'is lossless really lossless' etc etc.

Lossless vs. Redbook tests?

Reply #4 –
To be a little less terse.

It is not necessary to do listening tests to prove that losslessly encoded files are sonically identical to WAV/AIFF files. If a lossless file is properly decompressed, the output can be shown to be bit-identical to the original uncompressed file, by a simple file compare routine.

If (and it is rather a big if) it has been shown that there is a difference in the output from playing a lossless file, compared with a WAV/AIFF, then it must be because of some difference in the playback chain of the two files, or to some brokenness in the decoder. This might possibly include some (drum-roll) jitter, I guess, but I don't know enough about that to know if it's possible that this could arise because the different playback route, or how much there needs to be to be audible. But even if it's there, it's not produced by the file format.

I am no expert, but I do know that in the digital world there are some things that don't need tests, because they are Mathematics, and therefore certain.

Oblig xkcd: http://xkcd.com/263/

Lossless vs. Redbook tests?

Reply #5 –
WTF?  No need to be d_cks about it.  This may be a worn out topic for you, but I'm new to using a computer as a music playback device.  I searched for lossless in the test threads and came back with no hard data. 

If it is such a bother to help DON'T POST!  Shrug your shoulders and move to the next thread that interests you.

EDIT:  this response is directed at the first 4 posts.

Lossless vs. Redbook tests?

Reply #6 –
We respond because this is a topic which has been done to death and we do not wish anyone to trip over your (second hand) claims and give them any credulity whatsoever. Once the audio data is in PCM (i.e. from WAV or decoded lossless) the playback mechanism should be identical, therefore no different.
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

Lossless vs. Redbook tests?

Reply #7 –
Quote from: MichaelW on
If (and it is rather a big if) it has been shown that there is a difference in the output from playing a lossless file, compared with a WAV/AIFF, then it must be because of some difference in the playback chain of the two files, or to some brokenness in the decoder. This might possibly include some (drum-roll) jitter, I guess, but I don't know enough about that to know if it's possible that this could arise because the different playback route, or how much there needs to be to be audible. But even if it's there, it's not produced by the file format.


That's where the speculation is headed in the thread on the other site, that the data is all there bit for bit, but there may be something happening in the decompression of the file.

Lossless vs. Redbook tests?

Reply #8 –
If "the data is all there bit for bit" then how can there be any difference? It is (surely) already decoded?
lossyWAV -q X -a 4 -s h -A --feedback 2 --limit 15848 --scale 0.5 | FLAC -5 -e -p -b 512 -P=4096 -S- (having set foobar to output 24-bit PCM; scaling by 0.5 gives the ANS headroom to work)

Lossless vs. Redbook tests?

Reply #9 –
Quote from: Nick.C on
We respond because this is a topic which has been done to death and we do not wish anyone to trip over your (second hand) claims and give them any credulity whatsoever. Once the audio data is in PCM (i.e. from WAV or decoded lossless) the playback mechanism should be identical, therefore no different.


I'm making no claims, first or second hand.  I'm not looking for "should be identical" as an answer.  This is an objectivist site, and in that spirit, I just need some links to some studies and we can end this quickly.

Lossless vs. Redbook tests?

Reply #11 –
Quote from: AV-OCD on
This is an objectivist site, and in that spirit, I just need some links to some studies and we can end this quickly.

No need for any studies; you can test this yourself:

1) Download foobar2000 (if you don't already have it)
2) Add foo_bitcompare.dll to the components directory
3) Select a WAV and encode to a lossless (e.g. FLAC) file
4) Select both the WAV and the lossless (e.g. FLAC) track > Right Click > Utils > Bit-Compare Tracks ...

The result:

Quote
No differences in decoded data found.

End of story.

C.

[EDIT2: Changed it round to start with a WAV - as that makes more sense]
PC = TAK + LossyWAV  ::  Portable = Opus (130)

Lossless vs. Redbook tests?

Reply #12 –
Quote from: AV-OCD on
WTF?  No need to be d_cks about it.  This may be a worn out topic for you, but I'm new to using a computer as a music playback device.  I searched for lossless in the test threads and came back with no hard data. 

If it is such a bother to help DON'T POST!  Shrug your shoulders and move to the next thread that interests you.

EDIT:  this response is directed at the first 4 posts.


As people have already said, yes this has been done to death.

Typing "Is lossless really lossless" brings back many results about what you're asking.

http://www.google.com/custom?domains=hydro...D%3A1&hl=en

That's why people are being short with you.

Lossless vs. Redbook tests?

Reply #13 –
WAV1 > FLAC > WAV2.

WAV1=WAV2.

So unless you think that identical data stored on different (arbitrary!) parts of a HDD can sound different, those two are the same. If they are not, forget PC audio: either you are mad, or your PC is so broken that it probably can't even run Word.


FLAC > sound card uses the same decode routine as FLAC > WAV, but "running in real time" (rather than much faster than real time). So the same bits get transferred to the sound card in both cases. In most architectures the clock is the same in both cases, so jitter differences are usually unlikely.


FLAC > sound card clearly has slightly different processes running in the PC at the instant of playback to WAV > sound card. So, do these different processes affect the sound output?

Well, on really really bad old on-board sound cards, you can sometimes hear what the CPU is doing. Literally. You can hear when it's idle, when it's not doing much, and when it's working flat-out. It creates different sounds, distortion, or interference on the audio output.

However, even in that case, the difference between FLAC > sound card vs WAV > sound card is non-existent - both processes are such CPU-light operations that it doesn't matter.

If there is a PC system where there's some audible difference between FLAC > sound card and WAV > sound card, then launching, say, internet explorer (or whatever) would cause the audio playback to fall apart, because the CPU load and processes involved are dramatically greater. In reality, this simply does not happen!

As CPU loads 1000x greater than the difference between FLAC > sound card vs WAV > sound card do not cause any audible difference on any half-decent PC from the last ten years, it has to be concluded that anyone who claims to hear a difference between FLAC > sound card and WAV > sound card either has a broken PC/configuration, or is imagining it. To rule out the latter, double-blind tests are required. If these are passed, then the former can be debugged.


I wouldn't dismiss these claims lightly - of course lossless is lossless and sounds the same. However, it is possible to use two different PC audio players, route the audio differently within the OS without realising it, and hence actually send completely different audio from FLAC vs from WAV, even though the file contents are identical.

A few years ago it was quite common for people to say "WAV sounds different from CD" - where what they actually meant was "This badly ripped WAV file played back through my SB16 sounds different from the CD played on the in-PC CD-ROM drive connected through the analogue loop-through on my sound card, with the sound card mixer settings adjusted to be different for the two devices".

Far be it from me to suggest that the kind of people who believe in directional cables and non-lossless lossless formats are also the kind of people who would make such a basic mistake

Cheers,
David.

Lossless vs. Redbook tests?

Reply #14 –
All computer ABX protocols (that are used here, anyway) decode to WAV before the test even begins. So if a difference did exist, neither foobar2000 nor PCABX would be able to detect it. You'd need to write your own ABX setup with on-the-fly decoding to theoretically test that sort of thing.  That alone is a significant barrier from testing this. I think a lot of the replies here are a off base in suggesting foo_compare etc because of that.

More generally, once you start questioning "bits is bits", your entire audio epistemology breaks down, and you start having to admit that all sorts of things could work (cryo cable, power cable, Shakti stones...). And, of course, computer ABX testing becomes much less meaningful. For instance, maybe the fact that one has to write a custom ABX app to decode lossless on the fly means that it will sound different than a "real" music player... that's the sort of thinking behind XXHighEnd, the Memory Player, etc. Which all revolve around finding new and inventive ways of doing the exact same thing - copying data around in memory - because that's literally the only place the audible difference could exist, if it did. (It's also the only thing application developers have any real control over, which is only the latest in a long line of useless audiophile tweaks that only came about because they were the only tweaks that were possible.)

That said, you've gotta call that commenter's bluff when he's blowing steam about hearing about his "very knowledgeable engineer" friends confirming all this. If they can't cite the logical numerical evidence - and moreover, show that the evidence is a necessary and sufficient condition for the perceived distortion, rather than some random difference they pulled out of thin air that has no importance whatsoever - cut their f*cking balls off. People like them give engineers a bad name.

I myself prefer to think that even if there was a difference  between lossless formats, it is utterly counterproductive to my  well-being to actually believe it. In addition, of course, to being  thoroughly convinced that said difference doesn't exist in the first  place. There's a certain amount of axiom involved with all of this, and  short of exhaustive multi-thousand-trial ABX tests on the subject, on  HA, "bits is bits" is more or less axiomatic. Once that breaks, we  might as well go home. Fortunately we have an almost inexhaustible  amount of evidence, both numerical and otherwise, to back up (but not  necessarily prove) the axiom. The audiophool camp has a substantially poorer body of evidence.

Lossless vs. Redbook tests?

Reply #15 –
To simplify comparing a Lossless + uncompressed/CD file with a computer (binary comparison) is the same as using your ears to perform a blind listening test only it's 100% accurate.

Lossless vs. Redbook tests?

Reply #16 –
Quote from: AV-OCD on
Quote from: MichaelW on
If (and it is rather a big if) it has been shown that there is a difference in the output from playing a lossless file, compared with a WAV/AIFF, then it must be because of some difference in the playback chain of the two files, or to some brokenness in the decoder. This might possibly include some (drum-roll) jitter, I guess, but I don't know enough about that to know if it's possible that this could arise because the different playback route, or how much there needs to be to be audible. But even if it's there, it's not produced by the file format.


That's where the speculation is headed in the thread on the other site, that the data is all there bit for bit, but there may be something happening in the decompression of the file.


(shrug) tell them to bit-compare the decompressed file to the original.  It's easy to do with a file-comparison tool.  They'll be bit-identical unless something is very wrong.

You might also ask the 'speculators', who I'm guessing don't have much of a clue what they're talking about,  to put up or shut up , regarding these supposed 'engineers' and their supposed findings.  Ask for links and references.


Lossless vs. Redbook tests?

Reply #17 –
Quote
"I know of some very knowledgeable engineers who have measured differences between Apple Lossless files and AIFF files. They measured the files, once a sonic difference was heard in an A/B comparison, in the first attempt to quantify the difference. The measurements were carried out on several computers with some of the best test equipment available."


Then someone probably adjusted the gain (eg replaygain) of the lossless file.

Lossless vs. Redbook tests?

Reply #18 –
Quote from: uart on
Then someone probably adjusted the gain (eg replaygain) of the lossless file.

I would be very careful to choose my words better.  This could be horribly misleading.

EDIT:
What do I mean by this?  Let me first apologize for not taking the time to tell you in the first place.

uart, you should avoid giving the impression that soundcheck/replaygain somehow changes the audio data.  Saying that the gain of the "file" has been adjusted leaves room for this conclusion.  It is far better to make mention of a tag in the file that is not part of the audio data.  There are still people out there who are afraid that RG/Soundcheck does the same thing as the destructive process of peak normalization that is sometimes found in ripping programs.  We don't need to feed their fear.

Lossless vs. Redbook tests?

Reply #19 –
Quote from: AV-OCD on
"I know of some very knowledgeable engineers who have measured differences between Apple Lossless files and AIFF files. They measured the files, once a sonic difference was heard in an A/B comparison, in the first attempt to quantify the difference. The measurements were carried out on several computers with some of the best test equipment available."

As a "knowledgeable engineer" myself -- one who is actually firmly grounded in an objectivist reality, unlike many others -- I disagree with any test in which a raw audio file is demonstrated to be perceived any differently from that of its correctly-encoded lossless counterpart based on scientific grounds alone. Prior to performing a listening test, it should have been determined whether the audio data of the Apple Lossless file could be decompressed correctly (meaning that the audio data, once decompressed, is identical to the audio data of the original uncompressed AIFF). If the audio data matches bit-for-bit, any further testing is unnecessary: if the decoder can decode the stream correctly at a rate greater than real time, it can decode the stream correctly at any given rate. If the audio data does not match bit-for-bit, any further testing is meaningless with respect to the original and appropriate goal of the test: to determine whether audible differences can be discerned between a correctly-encoded lossless file and the original uncompressed file.

I wouldn't begin questioning the laurels of lossless encoding just because it's claimed that a group of undefined, unknown "engineers" did some sort of vague testing that yielded a strange result. Engineers make a lot of claims and do a lot of bizarre things, but the title certainly doesn't grant them (or me) the slightest bit of immunity.

Lossless vs. Redbook tests?

Reply #20 –
Quote from: greynol on
Quote from: uart on
Then someone probably adjusted the gain (eg replaygain) of the lossless file.

I would be very careful to choose my words better.  This could be horribly misleading.


Can you explain why? Is it because apple lossless doesn't use replay gain or some other reason. Change replaygain to soundcheck if you like. I'm talking about two "identical" files sounding different simply because one has meta-data instructing the playback software to modify the volume and the other doesn't. That seems like a simple enough proposition to me.

Lossless vs. Redbook tests?

Reply #21 –
The words imply that one of the files is lossless, not both. Is that what you mean, greynol?

Lossless vs. Redbook tests?

Reply #22 –
Quote from: gnypp45 on
The words imply that one of the files are lossless, not both. Is that what you mean, greynol?


Oh yeah I see what he means now.

Ok I meant the file encoded with the apple lossless codec. hehe

Lossless vs. Redbook tests?

Reply #23 –
BTW I don't think we should dismiss AV-OCD so quickly, I just did some tests that seem to support his idea.

Here’s what I did. I wrote a really small text file in notepad and then made a compressed copy with Winzip. I then extracted the file from Winzip and checked that it was bit identical with the original. However when I read the text from the second file I found it conveyed a completely different meaning, despite containing the exact same text (and spacing and font etc). Can anyone else please verify this incredible result, I've cut and pasted the contents of the two files exactly below.

This is the content of the first text file (uncompressed) copied and pasted exactly.

"AV-OCD is a very knowledgeable fellow."

Notice that in this first file the meaning is pretty much exactly literal.

Now compare the compressed file!! Here are it's contents copied and pasted exactly.

"AV-OCD is a very knowledgeable fellow."

Can you pick the difference in meaning. It’s totally amazing, though all the letters and words are identical you can clearly see that the meaning now has totally changed. When you read it you no longer get the literal meaning but instead it’s somehow saying to you that "AV-OCD is a total noob". This is totally amazing, I never would have expected it. Does everyone else get this strange new meaning form the text copied from the zipped file?

Lossless vs. Redbook tests?

Reply #24 –
Quote from: uart on
Can you pick the difference in meaning. It’s totally amazing, though all the letters and words are identical you can clearly see that the meaning now has totally changed. When you read it you no longer get the literal meaning but instead it’s somehow saying to you that "AV-OCD is a total noob". This is totally amazing, I never would have expected it. Does everyone else get this strange new meaning form the text copied from the zipped file?


Is it necessary to be so insulting?  Honestly, at least many audiophile forums are inviting to new members.  If hydrogenaudio wants to spread their word of objectivity, they're more likely to do it if people try to actually help the "noobs."  Are you trying to drive out the people trickling over here from other forums?

It would have taken 30 seconds for any of the people replying to this thread to run a search and post a link to an existing thread on this site.  I realize that the same questions come up over and over again, but that's the same on every forum in the world.