Apollo: improved MP3 decoder Options

[Speek]

From the Apollo website:

Apollo 37zh and sound quality (2004/11/29)

Download Apollo 37zh with some sound quality improvements in the MPEG decoder.

The sound quality of the decoder has improved quite a lot during the last three releases. To prove this, I measured the difference of Apollo 37zh 24-bit output to the MPEG-1 audio layer 3 compliance test reference signal (ISO/IEC 11172-4). More info on the test and results for a bunch of other players (including an old version of Apollo) can be found here. The results for Apollo 37zh 24-bit output are RMS level 6.835×10−9  and maximum difference 1.192×10−7.

For comparison, the same values for the latest MAD decoder (MPEG Audio Decoder 0.15.1b) built with accuracy optimizations enabled and 24-bit output without dithering are RMS level 8.804×10−8 and maximum difference 4.768×10−7. The difference is quite obvious and favorable to Apollo: RMS level is over ten times the one produced by Apollo 37zh and the maximum difference is four times the one by Apollo (interestingly the results for MAD 0.11.4b seem to be somewhat better than for the latest one but they are still worse than Apollo's). Actually, the maximum difference of Apollo's output is the smallest possible deviation in 24-bit data, the only smaller possible value would be zero. This means that at any time, Apollo's output is either exactly the same as the reference output or differs from it by the smallest unit expressible in 24-bit data. Of course, this only goes for the provided test signal, but it should give some picture about the sound quality.

[rjamorim]

Very interesting.

I wonder if the developer wouldn't be willing to release the decoder on an open source license less restrictive than MAD's. it might be a good opportunity to replace foobar's problematic mpglib.

[music_man_mpc]

mpglib is problematic? How so?

[rjamorim]

http://www.hydrogenaudio.org/forums/index....ndpost&p=222284

http://www.hydrogenaudio.org/forums/index....ndpost&p=222529

One of the issues reported was later fixed by the same guy that reported it, but there seems to be more.

This post has been edited by rjamorim: Dec 13 2004, 01:46

[metaller]

The results for Apollo 37zh 24-bit output are RMS level 6.835×10−9  and maximum difference 1.192×10−7.

Good, but not the best. The results for mpadec (and mpg123, but not mpglib) are RMS level 2.28x10-9 and max difference 1.192x10-7.

[john33]

Good, but not the best. The results for mpadec (and mpg123, but not mpglib) are RMS level 2.28x10-9 and max difference 1.192x10-7.

If anyone wants to take a look at mpadec, you can d/l a win32 compile from: http://homepage.ntlworld.com/jfe1205/mpadec.zip. It's a command line decoder.