Hacker News new | ask | show | jobs
by phendrenad2 757 days ago
The non-AI version is better: Using multiple microphones, you can use time-of-flight to isolate multiple voices in a crowd, and separate them into distinct, clear audio tracks.
1 comments

This does make use of multiple microphones and timing between sound waves arriving. To my understanding, the machine learning part allows it to work even as the speaker/wearer move relative to each other, such as if you turn your head.