Pitch tracking is the main thing I'm worried about at the moment.
Reading through forums here fiddle~ seemed to be a good choice, although I've also seen posts describing sigmund~ as a related and more powerful tool.
My problem is I can't find the PC versions of either, Miller Puckette's page for these externals directs to the above link for both.
I'm starting to think maybe I should try doing the port myself, is it a complicated process? It's not something I've ever tried before.
[pitch~] is what i always use, it's based on [fiddle~].
Just looking at [sigmund~] now (which i have, but haven't really used); I think the tracking feature is the main difference, although i might be wrong. The help file demonstrates much better resynthesis than the i've seen before.
And it's more max5 friendly, with attributes and whatnot.
Regarding the port, it looks like you would have to contact Ted Apel etc for the source code as it isn't hosted anywhere. It isn't hard in general, but pitch tracking externals themselves are complicated. Alternatively you could get the PD code (which i assume is available somewhere) and do it from that. There's a bit more to do in that case of course, as you have to port pd to max..
> Is there something that sigmund~ does that is not addressed by these?
Yes - not only does sigmund~ provide sinusoidal tracking and a couple of other new features, it is also features a much improved sinusoidal detection algorithm. Miller wrote sigmund~ specifically as an improvement upon fiddle.
A very basic explanation is that each peak is subtracted from the analysis (in order of amplitude - largest first) - because each sinusoid excites every band in the FFT to some extent, this process results in a reduction of the error introduced from nearby large peaks. Given a better set of values for the amplitude and freq of each peak the fitting algorithm performs better.
In terms of getting the code - the PD source has #defs for Max or PD compilation, so you can actually compile directly from the PD source file...