The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Acoustic Word Embeddings for Low Resource Speech Processing with Herman Kamper - TWiML Talk #191
Oct 16, 2018
Chapters
Transcript
Episode notes
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Introduction
00:00 • 2min
What Is the Best Thing About Ondaba?
01:58 • 2min
Is TTI Chicago Affiliated With Another School?
03:40 • 2min
How to Do Speech Processing When You Don't Have a Lot of Resources?
05:41 • 3min
Are We Still Able to Operate in the Statistical Domain?
08:57 • 3min
Is There a Pendulum in Machine Learning?
11:39 • 3min
Is There a Difference Between the Zero Resource Speech Recognition Community and Cognitive Science?
14:14 • 4min
Zero Resource Speech Recognition - What's the Difference?
17:48 • 4min
Is That Right?
21:40 • 3min
How Do You Choose a Sliding Window?
24:35 • 3min
Is the Language Universal?
27:53 • 3min
Is There a Difference Between Zulu and English?
30:36 • 3min
Cod Switching - An Interesting Research Area for Low Resource Speech Processing
33:09 • 5min
What You Want to Do Is Go From Speech to Text
38:32 • 2min
Yo Ocised, That's a Great Usecase
40:09 • 3min
Do You Know What You're Talking About?
42:40 • 3min
Is There an Iterative Cycle?
45:37 • 3min
Ind Computer Programming - Is That a Difficult Piece?
48:58 • 3min
Gibs Sampling Is a Sperical Technique, i.e.
51:53 • 5min
Machine Learning in Africa - What's Next?
57:08 • 5min