IIRC - one audio channel is used for the speaker, and the other is a series of beeps which map to facial movements:
https://makezine.com/projects/chippy-ruxpin/
Back when I thought it might be a fun side hustle to make cool youtube videos (long ago put to bed), I thought about making videos of Ruxpin singing death metal and stuff.
It's been years since it seemed like it was worthwhile to make Youtube videos... And the situation is getting worse for creators year over year.