I have some bits of recorded speech. I want to identify unknown speakers by comparing their speech with bits of recorded speech of known speakers (I think I know who the unknown speakers are.) I think this is called a voiceprint. I've analyzed spectra (essentially fourier transforms) with Audacity but I haven't been able to detect similarities by eye. Perhaps I could export Audacity's spectral analyses in some numerical format and work up a nearly-unique signature, but I'd prefer to use what someone else has written.

Anyone know?

I don't know of any. But, by eye their are instrument tuners in Linux that show more waves\* than Audio City I've not gone through installing them recently (trying by ear) so sorry I don't have a name for the one in particular I'm thinking...

A quick search for 'voice' or 'speech recognition in Linux' could help and in which distro of Linux?

Wikipedia has an article that might be useful:

https://en.wikipedia.org/wiki/Speaker_recognition

1 members found this post helpful.

For acoustic analysis of speech you could try wavesurfer or praat. FWIW, I prefer the former.
jdk