[wplug] MP3 or OGG-Vorbis recovery tool?

[Vanco, Don]

> -----Original Message-----
> From: Benjamin Slavin [mailto:bslavin_list at wavecrazy.net] 
> Sent: Monday, October 27, 2003 2:49 PM
> To: wplug at wplug.org
> Subject: Re: [wplug] MP3 or OGG-Vorbis recovery tool?
> 
> 
> I understand what you're saying, but still believe that it would be
> possible (if computationally infeasible) to reverse the process.
> 
> Your JPEG example is one I had considered before my message, and I had
> come to the conclusion that the file, when saved (even at 100%) is not
> identical to the original image (which could be, say, PNG or TIFF). I
> just ran a test, and even the color was changed there. The 
> problem I see
> is that it's identical to re-encoding a decoded MP3.
> 
> It is highly possible (and somewhat probable) that there's header
> information in an MP3 file which would be difficult to recreate the
> second time around. If not, perhaps a "watermarking" feature could be
> used to add the information to the waveform in a way that would not
> affect the audio quality, but would allow for a fast re-compression.
> This would require tools to be specifically written, but in 
> light of my
> current situation, they'd be highly valuable.
> 
> --Ben
> 
> On Mon, 2003-10-27 at 14:26, Bill Moran wrote:
> [...]
> > Not really.  .wav files are made by sampling the waveform at fixed
> > intervals (44khz, right?), whereas mp3s use more complex mathematics
> > to store the data.  I don't have a great understanding of 
> how the mp3
> > math works, but I've been lead to believe that it's similar 
> in concept
> > to how jpeg files store images.  If that is correct, here 
> is what your
> > circumstance is:
> > 
> > You start with a .wav (whether it's straigh from a CD, or 
> in a .wav file,
> > it's in the same format) and you convert it to mp3 by 
> changing the data
> > format and discarding some details that you'd never hear 
> anyway, this
> > results in excellent compression.  Now you convert the mp3 
> back to .wav
> > format.  The .wav is exactly the same waveform representation as the
> > mp3, but is not the same as the original .wav.
> > 
> > At this point, logic would seem to tell you that if you 
> saved the second
> > .wav as an mp3, but set the quality to maximum, you would 
> get an exact
> > duplicate with the same compression ratio ... however, in 
> the case of
> > jpeg files, this is not the case.  If you follow the same 
> process (going
> > from bmp to jpg and back) you get an image of equal 
> quality, but much
> > poorer compression.  If you compress the image any further, you will
> > lose _some_ quality.
> > 
> > Don't ask me why this is.  It seems to go against all the 
> mathematics I
> > understand, but I would assume the case is similar with 
> mp3s, since they
> > are basically jpegs for sound.
> > 
> > I think what happens is this:  In the case of an mp3, the compressed
> > waveform itself is not the only data stored, also stored is 
> information
> > about what was done to make the waveform compress so well (which
> > information was discarded) but the .wav is not able to 
> store that second
> > bit of information, so it's lost when you repeat the process.
	
	Both issues (image and audio) relate to interpolation.  There's some
point where data location "X" has to be best guessed by the encoding /
decoding process and the complexity of the algorythms used to fill in those
data points.