Vincent Lostanlen
Vincent Lostanlen
Thank you for your kind words, @bmcfee. Come to think of it again, it's true that running YIN under the hood and displaying it automatically is going too far. And...
> I was imagining that the melody values would just provide a displacement of the existing wave plot, which is a single fill_between artist. We might want to break that...
I understand now. It's numerically linear but visually logarithmic, thanks to the `cqt_*` decorators. Neat!
Proof of concept below: 
@bmcfee Here is the code: https://gist.github.com/lostanlen/6abbfdf505df1779e7b67940e0631057
Thanks for sharing, @nullmightybofo . > The proposal in this issue is specifically about displacing waveform plots (like in waveshow) by f0 estimate, and the main challenge there is scaling...
OK! it's really no problem thanks!!
As part of the HAMR hackathon, I'm addressing this issue today. I'll try to make a spec this morning on which we could discuss, and then write the code in...
So @bmcfee sat together for a while and worked on a strategy for this. The plan is now to write a backend function to design a filter in the Fourier...
Probably relevant: https://github.com/KinWaiCheuk/nnAudio and Figure 11 of https://arxiv.org/abs/1912.12055