I'm attempting to write a patch that will in part automate the process of
transcribing relatively simple new agey music (a current project I've put off way too long). I'd like to know if fiddle has a higher degree of accuracy in detecting pitch in sound files sampled at 88.2k as opposed 44.1k. I know it said somewhere in the documentation that the interpolation of tabread4~ (what I'm playing through into fiddle) is far more accurate when the sound files are upsampled to 88.2k. This makes sense, of course, I'm just wondering that since I'm just playing the sound files back at "normal" speed if there is any other advantages, namely accurate pitch detection, to upsampling.
Since I'm working from a sound file and not a live performer, are there any other tricks to getting higher accuracy in pitch detection?
What are the limitations on the number of simultaneous independent fiddle objects? The reason I ask is that I'm thinking of dividing the register of the piano into band-limited regions, theoretically expanding the number of voices that can be detected simultaneously and thus be recomposited later. Or at least that's what I'm hoping to do.
I'm a little confused about the pitch output of fiddle's outlets. Currently I have one fiddle object declared thus:
fiddle 1024 3 20 0
This creates an object with six outlets. My understanding of the individual outlets is as follows: (Please correct me if I'm wrong)
outlet 1 cooked pitch (i.e., most prominent?) outlet 2 bang on attack outlet 3 raw pitch 1 (same basic pitch as cooked?) and assoc. amp outlet 4 raw pitch 2 and assoc. amp outlet 5 raw pitch 3 and assoc. amp outlet 6 overall amp
Ben,