Speaker recognition

I have audio I wish to identify the speakers in. I know they’re in a subset of about 10 possible speakers, have known samples of their speech. Is there a way to compare unknown samples to known samples to identify the speaker?

Slackware, kernel 5.10.1, Audacity 2.4.2, I compiled it myself today.

“Speaker” as in “loudspeaker” or “public speaker”?

It’s virtually impossible for a computer program to determine “who” is speaking - perhaps some of the latest experimental AI can make a guess to choose between training sets.

I want to identify which person is speaking. People have worked on this; technology for it exists. A frequency spectrum from a Fourier transform would be a good start. Audacity gets some of this information. Perhaps an Audacity user has tried it.