Robin Schmidt

...hmm...well...with more such bumps, it sounds more like phase-modulation, which makes sense and is perhaps not that interesting...we'll see

> We'll probably have to require a high samplerate for low frequency instruments, but it won't be uncommon to have to deal with low frequency instruments where the harmonic spacing...

> I thought this stuff was easy because all the FFT programs I've used, if you isolate a band of frequencies in the FFT program and then do a time...

as said - spectrogram resynthesis actually works already (well, not quite exactly the way i want it to, but still). with an appropriate gui, it should be possible to select...

did you watch these 3 videos? at the beginning of the 1st, he mentions the STFT - that's just another word for spectrogram: https://www.coursera.org/lecture/audio-signal-processing/sinusoidal-model-1-gjiP7 ...it's a lot of math, but...

> So is decent resynthesis easier that sinusood models? If so, why don't I use that for harmonic extraction? well, (re)synthesizing a signal from a spectrogram is easier...or let's say,...

i'm not sure, if i understand completely what you are suggesting. do you mean, we should forget about time-domain subtraction to obtain the residual and instead do the subtraction in...

this is how a typical short-time spectrum looks like:  do you want to zero out all the lobes and just leave that little stuff in between untouched? or do...

judging from the spectrogram plots, here, the window size must be at least 3 or 4 times as large as would be strictly necessarry for sinusoidal parameter estimation (the empty...

there are lots of linear algebra libraries around, and i'm comparing features, licensing issues etc. ...as well as the option of rolling my own code for this problem (actually maybe...