-
Notifications
You must be signed in to change notification settings - Fork 149
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
MP3: Mismatch with max delta from 0.002 to 2.0 #78
Comments
The chunk I've tested covers bitrates from 32kbit/s to 320kbit/s and a wide range of encoders. List of distinct encoders from ID3 metadata
|
Hi @Shnatsel, This may take me a while to get to. Do you know if these also all fail with |
20 out of the 88 files in the archive have significant divergences from both |
The file sharing service seems to be having issues with uploads, so here's a list: Mismatches with both ffmpeg and mpg123 (listed max delta is against mpg123)
|
File sharing is back up, so here's the archive with only the files that fail in both mpg123 and ffmpeg: |
Would it make sense to re-run the test on the MP3 corpus with the new FFT implementation? Edit: nevermind, I'll get a bigger corpus and re-run. I am particularly eager to get the MP3 decoder fully working because Symphonia provides the only pure-Rust, 100% safe MP3 decoder, and I'd like to make it the default in Rodio which currently relies on C code. |
Only the first packet should contain these tags. If they appear elsewhere in the stream they will decode to silence. Addresses #78.
Hi @Shnatsel, No, you don't need to re-run the corpus. The FFT is not used by the MP3 decoder since it has no benefit in terms of accuracy or performance. I went through all the files and they are all either now fixed, or not a bug. Details below. I'm hoping that 0.5 can be released in 2 weeks or so. The only outstanding issue I really want to solve before then is #96. I also want to get #66 merged for the release as well. This file had multiple LAME Xing/Info packets which were rightly discarded by
These files have a VBRI packet which was correctly being decoded as silence, but should've been consumed by the demuxer. This was actually on my todo list, I just didn't have a sample file. It is now implemented and these files are fixed.
The following files are corrupted and produce audible artifacts in both
The following files have junk data after some packets that cause
This file is perfectly okay, but
The rest of the files were fixed when I added proper support for gapless playback a few weeks ago. |
These files show max delta ranging from 0.002 to 2.0 when compared against
ffmpeg
withsymphonia-check
.Tested on commit 33f0701 (latest as of this writing).
As we have learned in #72, ffmpeg has divergences with other decoders, so running the check on these files against other decoders seems prudent.
The files are from the Pony Music Archive corpus and are free to redistribute. I have run the check on half of it because I don't have enough disk space for the entire thing. This amounts to roughly 90Gb of MP3 (12831 individual files).
The text was updated successfully, but these errors were encountered: