Spear also suports text outputs, and i find that very useful as well.
You can set sensitivity and threshold when you do the analyze.
Spear’s default works fine but you can tweek them to get what you want.
i made spectral additive synth and spectral comb filter with SDIF which both sounded very interesting. SDIF is kinda fun…
Getting small files and getting accurate files are opposite problems.
For accurate files, the first thing is the sound itself. Sounds with less reverb and less noise are good. Saxophone seems to work well, but shakuhachi is really hard. When analyzing in Spear, you can increase the frequency resolution, and change some of the partial id parameters.
Smaller files are a different story. I usually go to the Edit menu and "Select partials below threshold…" I select everything below -60db or so and delete it. That thins out the file by taking out stuff you probably won’t notice.
This is all as much art as science. If you want a really good result, expect to experiment a bit.