Hi

I've found in another forum (btarg.com.ar) a guy that said he's coursing soudn engineering at the university and has a teacher that said that FLAC, sometimes, is not 100% loseless and lose some bytes and recommend APE because he said that is 100% loseless.

Could it be real ? Somebody listened something in this way ?

Thanks in advance

C

Ok in this case it's not exactly rocket science. Flac a Wav then decompress it again and compare that to the original. You only have to be careful to compare the soundcontent because sometimes wav's have additional metainfo that may be different from the decompressed wav in which case CRC's would not match.

http://flac.sourceforge.net/faq.html#tools__wave_flac_wave

That teacher obviously isn't as smart as he thinks he is.

Yes, its what I think...

C

Maybe it would be wise for him to attend some masterclass

By masterclass you subscribe here, right?

People are so used to lossy audio that even FLAC might get judged by the 'amount of bass' and 'brightness' in the sound. There are only two possibilities – either the codec works properly and produces bit-identical output, or is broken.

Why aren't there complaints about PK Zip being lossy?

I've noticed that my text files compressed with pkzip aren't as readable as the originals. Something about the line spacing that gets lost in the compression.

Of course!

I knew I didn't need glasses!

Modulo bugs FLAC is lossless. See http://flac.sourceforge.net/format.html for a description. However I do not know of a fully written out mathematical proof anywhere. Mathematically the fact that every test vector that anyone has ever thrown at it has come out lossless doesn't actually prove that it's lossless.

Both APE and FLAC work in the essentially the same way exploiting the fact that sequential samples in an audio signal usually do not differ very much. They using a predictor to try and generate the signal from a function then encode a residual (the difference between the predicted and original signal) using rice codes.

I originally stated in this post that APE was closed source. APE was certainly originally closed source but it has been pointed out to me that the source was released sometime between 2002 and 2004. Unfortunately the author chose to release it under a license which made incorporation in to other projects and redistribution as part of things such as Linux distributions difficult.

True. Late Edsgar Dijkstra argued that programmers should provide the mathematical proof that their code is valid and always produce wanted result.

But that's too much to ask from a programmer. They are not all skilled mathematicians, since such proofs can be a real pain in the a.s

We are teached maths, and if doing strict coding, a function has to define the set of values for which it can work with, and the expected result that it generates.
On the other hand, not every bit of code can be represented with maths (how would you explain in maths printing a text on the screen?)

True, but printing to a screen doesn't involve any computation of numbers. Whereas all audio encoders do, therefore most of their mathematical algorithms can be proved.

Ultimately the only thing that a computer inherently understands IS numbers. How we encode other types of information, sounds, pictures, etc., is called programming.

Of course, but you don't code that unless you're using Assembly.

How we encode other types of information, sounds, pictures, etc., is called encoding.

Well, I didn't ask Josh for -V but it's there... You can be assured that all files made with the -V option are even more likely to be lossless.

Yes and in the way of worrying about cosmic rays changing memory during encoding operations lies madness

Is "Sound Engineering" really Engineering?
Based in the teacher's info. Not likely.