Is there no quality control within the MusicBrainz scan process?
I notice that MB is fingerprinting and passing obviously defective files; For Example:
2 sec duration & 11257 bitrate
22 sec duration & 2185 bitrate
Shouldn’t MB and/or SK summarily reject most files with faulty frames?
=====
Does Jaikoz, a much more complex program, have, or could it have, any pre-screening capability for corrupted files?
Im unclear how am I meant to know they are faulty?
Maybe you could start by restricting the SK “fix” to reasonable standards:
[i]The MP3 audio format lossy data compression. Audio quality improves with increasing bitrate.
32 kbit/s - generally acceptable only for speech
96 kbit/s - generally used for speech or low-quality streaming
128 or 160 kbit/s ? mid-range bitrate quality
192 kbit/s - a commonly used high-quality bitrate
320 kbit/s - highest level supported by MP3 standard[/i]
Seems strange to me that through the years of the developing mp3 specs, there is no utility which can id truncated audio and delete the files, which may be a contributing factor for short duration files with out-of-range bitrates.
How does the fingerprinting process fit into this puzzle? Can’t it make any determination regarding audio integrity?
I don’t really think this in scope as you’ve described it, okay you have a very low quality audio track, but if that is the only version you have do you really want SongKongKong to ignore or delete it.
When you have multiple recordings of the same tracks SongKongs Delete Duplicates task will delete the lower quality one and I do plan to provide a few more options for Delete Duplicates such as http://jthink.net:8081/browse/SONGKONG-387 so might fit in there.