Overview
Explore speech synthesis and text-to-speech generation in this comprehensive lecture from the CMU Low Resource NLP Bootcamp 2020. Delve into the history of unit selection, understand synthesis techniques, and learn about research tools and common datasets. Discover how to choose a voice, explore state-of-the-art synthesizers, and examine challenges in other languages. Investigate statistical parametric synthesis, neural farms, and the hard problems facing the field. Gain insights into unseen tests and engage with thought-provoking questions in this in-depth exploration of speech synthesis technology.
Syllabus
Intro
Agenda
History of Unit Selection
Understanding Synthesis
Research Tools
Choosing a Voice
Flight
Common data sets
The blizzard challenge
Testing
Stateoftheart
Other languages
Synthesizers
Unit Selection
Talking Clock
Just a Minute
Backchanneling
Statistical parametric synthesis
Neural farms
Nice sentences
The hard problems
Unseen tests
Questions
Taught by
Graham Neubig