Kat Black

Imagine searching a multilingual archive of text passages with a spoken question. The usual route transcribes the audio into words and then runs a text embedding system, adding a processing step and discarding whatever the words fail to capture, such as n... See more