Hello,
I posted a message on this a while back, but having done a little more playing with the feature, here are some comments and questions
FIRSTLY
According to my calculations, MJ, calculation of the BPM is usually wrong. Usually this difference is by about 5 or 6bpm's but occasionally can be anywhere between 0.5x and 2x the original BPM (and not necessarily _at_ those boundary values either.
My initial feeling is that MJ is taking the number of beats it detects in the music and simply divides by the lenght of the track. Is this approxmiately correct?
Is it not possible to take a "representative" sample of the music and take a BPM reading from that. Most often music will slow down towards the end, or have a funny middle or even start uncharacteristically which might make beat detection difficult.
Surely MJ can calculate a number of different BPM values by analysing different portions of the song and then compring these to see what makes the most sense.
I imagine that this would require a little sound analysis so that areas of similar intensity and so on were compared and no huge change in intensity occurred during each sample.
This would serve, as an immediate consequence, to slow down the analysis of each track.
An observation: tracks with higher intensities seem to have a more accurate BPM reading than those with lower intensities. However, I have yet to see a track that is not at least 5bpm lower than what it should be.
Incidentally, the software that I have written to test this is hardly complicated and requires you to simply push enter on the beat. A reverse metronome, if you like. It is accurate to within 2 seconds (which is not fantasticly accurate, but you get a reasonably accurate reading after 30 seconds or so...). All of 10 lines of code...
SECONDLY
Use a similar method to find a representative Intensity. Sometimes I get the feeling that the intensity is "averaged" too. I have Queen's "Bohemian Rhapsody" which starts off at 1, but then changes, mid point, to at least a 3 or 4, but MJ still keeps it as intensity 1.
I would like some sort of option to allow MJ a little more flexibility when dealing with Intensity to say well, what is the most representative intensity? To me, an Intensity of 1 is something to chill out to and absolutely nothing more should be entered into! Bohemian Rhapsody, on my scale, would be at 2 or 3, rather than a 1...
If there are reasons for why MJ is doing what it is doing, please let me know!
Author
Topic: BPM and Intensity calculations (Read 1349 times)