Help - Search - Members - Calendar
Full Version: floating-point MP3 decoders
Hydrogenaudio Forums > Lossy Audio Compression > MP3 > MP3 - General
Omion
I just ran into a file which, when run through my mp3packer program, shows up as different in Foobar's "bit-compare tracks". The odd thing is that the max. difference is -135dB, which neither 16- nor 24-bit decoding will be able to catch.

My question is: is there another free MP3 decoder which will decode to (at least 32-bit) float? I would like to check to see if the problem is in mp3packer, or if it's a rounding error in Foobar's decoder.


I honestly don't see how it could be either mp3packer or Foobar, though. Generally, if mp3packer messes something up, there are BIG differences in a few places in the file. This file shows a < -135dB difference all throughout the file. Foobar's MP3 decoder will have recieved the same data with both files, so even if it was a rounding error, it should be the same for both. Something weird is going on blink.gif
Garf
Do you have anything to reproduce it with?
Omion
The original file is here. You can put it through mp3packer or get the output file here. Foobar reports
CODE
INFO (foo_bitcompare) : first different sample found
INFO (foo_bitcompare) : differences found: 1056502 sample(s), starting at 1.85034e-002 second(s), peak: 1.788139e-007 at 9.373016 second(s), 2ch

This corresponds to a max of around -135dB.

Finding the differences in Audition shows white noise with amplitude approximating the amplitude of the original waveform.
Omion
Never mind!

I just tried the bit-compare on Foobar 0.9. It returns "no differences found"

I guess it was just a problem with the 0.8.3 MP3 decoder. Weird. No other files have done that. blink.gif
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.