Speech Processing - 5PMSPAR0

Goal(s)

This serie of lectures will cover the fundamentals of automatic speech processing including fundamentals of speech production and perception, acoustic phonetics, speech signal analysis and transformation, automatic speech recognition, Text-to-speech synthesis, statistical voice conversion.

Content(s)

This serie of lectures will cover the fundamentals of automatic speech processing:

• Introduction to speech science (speech production/perception, acoustic phonetics)

• Speech signal analysis (STFT, cepstral analysis, pitch detection, voice transformation)

• Automatic speech recognition (template matching, HMM-based approach, neural approaches including LSTM, CTC, and seq2seq+attention model)

• Voice conversion (using Gaussian mixture regression and neural approaches)

• Text-to-speech synthesis (from unit selection to neural TTS)

Basics of digital signal processing, machine learning