AVS Forum banner
1 - 14 of 14 Posts

·
Registered
Joined
·
2,181 Posts
Discussion Starter · #1 ·
A while back when I set up my music server, I read what I could find, here and elsewhere, about file formats. I selected FLAC. It is claimed to be lossless.

Now, I have some experience with file compression. TAR, ZIP, etc. I know for a fact the uncompressed file is absolutely bit for bit identical to the original. Granted, this is not a real-time process. It reads and writes by the OS drivers when allowed.

I have seen several discussions that some folks hear a difference between a WAV and FLAC file. I will not dispute what "some folks" hear as I am not them. Real or imagined, one must respect their perception.

So, is there actually a difference in the bitstream as delivered to the DAC? Some have suggested it has to do with metadata. This I do not believe unless you were streaming over a bad network, had a video running, maybe a spectrum analyzer, and were playing a game in the background all with run priorities not set on a 15 year old Celeron single core.

Some have suggested it is about processer capabilities to decode in real time. I can see this as a unlikely possibility, but does not defy the laws of physics considering the USB is not a real-time synchronous transport layer and who knows what Windows decides to do in the background. ( Give me True-64 RT!)

I could I guess see time domain issues depending on the unpacking, but I have to believe any decent CODEC has enough buffers to do the checksums and clock out a FIFO. So, maybe a really crappy CODEC?

So, my question is, is there any test suite that can send a WAVE and a FLAC file out and in and do a bit-comparison? That would prove the format is lossless. I am not sure how I would measure the jitter in the USB stream without some fancy test equipment.

Can anyone point to OBJECTIVE tests of this nature? I am not at all interested in I think XXX sounds better.
 

·
Registered
Joined
·
378 Posts
Real or imagined, one must respect their perception.
A sighted test says a lot about perception of the listener and very little about de object tested.
This we know for a century, e.g. https://en.wikipedia.org/wiki/N-ray
Might it be that if somebody mistakes his perception for the properties of the DUT, he is making a pretty stupid attribution error?

Some have suggested it has to do with metadata.
Total nonsense.
Send meta data to a DAC e.g. album name or cover art and it will crash as a DAC converts digital audio to analog audio.
That would prove the format is lossless.
It is 2021, you still need a prove FLAC is lossless?
A simple test: take a WAV1, convert it to FLAC, covert FLAC to WAV2
Use any tool e.g. Foobar binary compare to compare WAV1 with WAV2
This will tell you that even in 2021 FLAC still is lossless.

A bit more complicated:
Convert a WAV to FLAC.
Play both and record the digital out.
Time align both recordings and do a null test only to find zero differences.

Wait a minute. We are still in the digital domain so because both are lossless there will be no difference.
But FLAC produces more jitter than WAV!
Nobody has ever proven this but the armchair engineers on the audiophile fora know for sure this is the case.
If you don't want to fork our $30k for a AP kit you can try:
Convert a WAV to FLAC.
Play both and record the analog out with a good ADC.
Load both in software allowing you to do null testing e.g. DeltaWave
"mitchco " did so years ago detecting zero differences. Unfortunately, his posts disappeared from CA The Well-Tempered Computer

You want to hear the difference?
Load the WAV and the FLAC in Foobar and use the Foobar ABX comparator
Can you find any ABX on the internet proving there is a difference between the two?
No, you can't because there isn't.

I'm afraid you are not stirring any pot but just wasting your time.
 

·
Registered
Joined
·
5,876 Posts
I have seen several discussions that some folks hear a difference between a WAV and FLAC file. I will not dispute what "some folks" hear as I am not them. Real or imagined, one must respect their perception.
You're a better person than I am. I wouldn't respect anyone that thinks there is a difference.

It is 2021, you still need a prove FLAC is lossless?
A simple test: take a WAV1, convert it to FLAC, covert FLAC to WAV2
Use any tool e.g. Foobar binary compare to compare WAV1 with WAV2
This will tell you that even in 2021 FLAC still is lossless.
Exactly, there is no difference. That's the whole point of "lossless".
 

·
Registered
Joined
·
140 Posts
WAV or FLAC. doesnt matter to me. Ive been messing around with rips and downloads for as long as the technology has been available. I dont like how it sounds. Now I dont have a multi kilo buck system like a Baetis or others but I do have a multi kilo buck disc reading and playback system with USB file read capability. Give me the polycarbonate disc every time.
 

·
Registered
Joined
·
5,876 Posts
WAV or FLAC. doesnt matter to me. Ive been messing around with rips and downloads for as long as the technology has been available. I dont like how it sounds. Now I dont have a multi kilo buck system like a Baetis or others but I do have a multi kilo buck disc reading and playback system with USB file read capability. Give me the polycarbonate disc every time.
I don't miss playing CDs at all. I like the convenience of playlists, album art, track name, year, etc. all on a nice display. The only use I have for a CD is ripping it as soon as I buy it. Then it just goes into storage.

Not to mention I don't have to worry about the center of the CD cracking, rot or scratches.
 

·
Registered
Joined
·
140 Posts
I don't miss playing CDs at all. I like the convenience of playlists, album art, track name, year, etc. all on a nice display. The only use I have for a CD is ripping it as soon as I buy it. Then it just goes into storage.

Not to mention I don't have to worry about the center of the CD cracking, rot or scratches.
Of course to each his own. Ive never had a "rotten" commercially made cd, or cracks or delamination. Had some black homemade cd's refuse to play. I had a Buddy Guy cd one time that would not play on 3 different player and transports I had but would play on a friends boom box.
 

·
Registered
Joined
·
2,181 Posts
Discussion Starter · #7 ·
I remember the rot hype. My first CD was Cats. Yes, I am one of the three who like it. Anyway, it was mid 70's and it works fine today.

I don't have a mega buck DAC either. Schiit Asgard.

I understand the file compressed, stored, uncompressed to a FILE would be identical. As I said, I am familiar with TAR, ZIP and the like. One bit off and you would know it. It is the not very real time, unknown buffering and what services may interrupt the best effort dump to the USB that may be what is being whining about. ( give us back firewire)

So, here Joe says he does not like how it sounds. Well, if that is the case, then something is not identical.

FOOBAR binary compare. Thanks. A tool I did not know of. I downloaded FOOBAR but have not managed to get the UI into a useable form, so got no further. WinAmp is the other with possibilities. Currently use Windows Media Server as the UI makes sense. Just drag stuff over to a temp list, no new list, no typing names no hassle.

No way will a segment decoded to analog, then digitized again will match bit for bit the very same segment again. Not buy the laws of physics. Timing won't match so the values will be different. Nyquist may have been right in theory, but the basic theorem did not include the filtering and artifacts produced.

So here we have some strongly in the "it's BS like magic cables" camp, and one who says he thinks CD's sound better.
I am in the "highly skeptical as I don't see why camp". Not proved yet.
 

·
Registered
Joined
·
5,876 Posts
No way will a segment decoded to analog, then digitized again will match bit for bit the very same segment again. Not buy the laws of physics. Timing won't match so the values will be different. Nyquist may have been right in theory, but the basic theorem did not include the filtering and artifacts produced.
What is this analog step you're talking about?
 

·
Registered
Joined
·
2,181 Posts
Discussion Starter · #9 ·
Reply to the suggesting for testing to play back and then digitize, comparing the original digital to the result.
 

·
Registered
Joined
·
984 Posts
Absolutely it could be the case that some particular piece of software on some particular piece of hardware could give better or worse results with FLAC versus WAV.

But that would boil down to some sort of bug.

Such bugs are not uncommon. Particularly in the multichannel domain. Some device might play 4.0 WAV but screw up 4.0 FLAC. Or vice versa.

But in principle, no, if everything's working correctly, there's no reason they should differ.

If they do differ due to a bug, either one could be better.

It might also be possible for either a FLAC encoder or decoder to have a bug in the actual codec, so that they don't give bit-exact results. But any such bug should be detected by a very basic test suite. Chances of an error like this occurring is minimal assuming a device is using a standard library.
 

·
Registered
Joined
·
13,455 Posts
I remember the rot hype. My first CD was Cats. Yes, I am one of the three who like it. Anyway, it was mid 70's and it works fine today.
Wait...what?? :unsure:
 
  • Haha
Reactions: Ratman

·
Registered
JVC NX5 at 140", Denon X4200W (5.1.2) with Axiom Audio speakers + Bass Shakers
Joined
·
3,448 Posts
The decoding of the FLAC happens well before the audio is sent out whenever transport to the DAC you are using.

The exact same signal is sent out of the software whether you are using WAV or FLAC or any other lossless audio format.
 

·
Registered
Joined
·
3,344 Posts
It might also be possible for either a FLAC encoder or decoder to have a bug in the actual codec, so that they don't give bit-exact results. But any such bug should be detected by a very basic test suite. Chances of an error like this occurring is minimal assuming a device is using a standard library.
The FLAC build process actually tests the binary it makes for this very problem. The build ships with a sample test file, and the file is encoded and decoded and the files compared to make sure the host system that built it is sane and works as expected.

Now, it's possible that other people using the FLAC code don't take that kind of care and end up with a buggy version, but the official versions are tested as are the versions you build yourself.
 

·
Super Moderator
JVC RS4500 | ST130 G4 135" | MRX 720 | MC303 MC152 | 6.1.4: B&W 802D3, 805D3, 702S2 | 4x15 IB Subs
Joined
·
12,345 Posts
I remember the rot hype. My first CD was Cats. Yes, I am one of the three who like it. Anyway, it was mid 70's and it works fine today.
What are you saying? The CDROM was invented in 1979 and the first CD was released in 1982.
 
  • Like
Reactions: Alan P
1 - 14 of 14 Posts
Top