Translating meeting audio with AI in real time

How do we translate other speakers’ audio in a meeting with AI? (Note: I use this with the other parties’ permission; running it through AI without consent may not be ethical.)

First we need to create a virtual microphone and feed the external audio into it. I use Loopback for this. (Note: it works at the OS level, so be mindful of the risks.)

With the flow set up in Loopback, we can send the audio destined for the headphones to both a virtual microphone and the headphones. From there you can point any translation/transcription tool at the virtual mic.