
Acoustic Word Embeddings for Low Resource Speech Processing with Herman Kamper - TWiML Talk #191
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
00:00
Yo Ocised, That's a Great Usecase
A model can do this, so you can fuddle around with the model and check the type of mistakes that the model makes. And in that way, we can actually learn something about owr, humans. A third motivation is, and this is actually a project that i'm working on, amd, with am, some of the colleagues here. M, it's axally a comm a project with the united nations. They're really interested in very, very specific key words. The data could be used to figure out what people are talking about in uganda.
Transcript
Play full episode