i looked at this thread about a month ago (as they say, it pertains to my interests), and was surprised nobody mentioned how poorly MC determines track BPMs. then when i looked at it just now, i was really surprised nobody brought it up. surely i can't be the only one who's seen this?
now i recognise that automagically determining a BPM isn't the easiest, and MC is right about half of the time (within a few BPM of actual). 25% of the time, it's on a ratio of the actual BPM. usually it's about 66% of the actual, or 125%, or 166%, or 50%, or something. those i was willing to call ok, figuring that they'd have complex rhythms. i took a look at the tracks in my collection that are somewhat, though, and they're not at all complex. i thought they would be drum'n'bass or jazz or something, but a lot are regular tracks or even something super-simple like techno.
anyways, the other 25% of the time, it's just plan wrong, and there's no identifiable pattern to it.
usually what i do is, after importing, i re-analyse the BPMs using a program called "mixmeister bpm analyzer". that program is hardly ever wrong, and if it is, it's within 1 BPM of the correct number.
so, i recognise this is probably not very high on the j. river priority list, but i am curious if anyone else has problems with inaccurate BPMs in the collection, and if j. river would consider taking a look at their algorithms for analysing this. (i wouldn't think of suggesting they copy what the mixmeister program does, but it is right 99% of the time...)
thanks!
micah