> One comment: that test case (the only one for which the reference
> decoded output is provided) is very simple. It doesn't test commonly
> used features like jstereo (or stereo), short blocks, scfsi,
> scalefac_scale and scalefactor pre-emphasis to name a few.
>
> So I would mention that if a decoder failes this test, it is quite
> bad, but even if it passes this test, it could still have many
> serious bugs (as shown by some of the other decoder tests).
Rob done a good job by spending time testing compliance of decoders (even if
I would like to see the result of the dist10 decoder), but as Mark
mentionned, it seems that this test is not enough. In the test package,
l3dec seems to be implicitely reffered as the reference decoder (althought
I'm asking myself what was used to produce the reference output).
So here is my suggestion (and also a lot of work to do):
*decoding the others test bistreams with the latest l3dec (I think it's
2.72)
*perhaps adding a 0-22k sine sweep, but in this case wich encoder can be
used to be totally objective?
*using those "reference" decoded streams to test the previously tested
decoders
If no one have time for this, I'll have the time for this task in september
Regards,
--
Gabriel Bouvigne - France
[EMAIL PROTECTED]
icq: 12138873
MP3' Tech: www.mp3-tech.org
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
