I tried my channel vocoder out on more signals, and I'm pleasantly surprised that it was able to make this out of this bassoon snippet:



And some talking: