[FR] Support "the method of reassignment" for estimating sharper frequencies in spectrograms

[kabocha-0]

ACK that visualization is still WIP; creating this issue as requested here. :)

Per Wikipedia:

The method of reassignment is a technique for sharpening a time-frequency representation (e.g. spectrogram or the short-time Fourier transform) by mapping the data to time-frequency coordinates that are nearer to the true region of support of the analyzed signal.

I'm most-familiar with this from Minimeters's "Sharp" and "Sharper" setting, vs. it's normal "Classic" spectrogram setting.

Reassignment isn't perfect in every scenario from my experience; sometimes I prefer to just use a normal non-"sharpened" spectrogram instead. But sometimes, it's a really nice force-multiplier for audio analysis, especially when working with bass-heavy stuff where the FFT buckets represent a huge perceptive range.

Here's Au5 talking about reassignment and comparing some spectrograms with/without it, for more context.

[wizenink]

Doing this would require computing 3 STFTs:

• Standard STFT
• Time-weighted STFT with time-displaced window
• Derivative-weighted STFT with the derivative of the window (approximating partial derivatives)

Would have to check how to implement it