Do-It-Yourself Speech Synthesis

Do-It-Yourself Speech Synthesis

This course is a hands-on course in which each student, working with a team of fellow students, will build a speech synthesis system in the voice of each individual student. In broad outline, the steps in the process of building a speech synthesis system for a particular voice are as follows:

Obtain a pronunciation dictionary

Record acoustic data

Train speaker-dependent speech recognizer

Align acoustic data to phonetic transcription

Obtain letter-to-sound rules

Select inventory of synthesis segments

Implement optimized-transition segment concatenation

Diagnose with speech mimic data

Test intelligibility with human listeners

Test quality preferences

Iterate segment selection and concatenation

Optional extra projects:

Pitch-preserving time-duration adjustment

Formant-preserving pitch adjustment

Amplitude adjustment

Articulatory analysis

New languages