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
Are We Still Able to Operate in the Statistical Domain?
A lot of speech recognition systems still operate in the statistical domain. People are trying to figure out what type of cus, what type of extra structure we need to build into our supervise models for it to actually learn language. If we can answer those questions, then that might tell us something about how and the catakof cus that human, human infants use to do this.
Transcript
Play full episode
