High-Fidelity Neural Phonetic Posteriorgrams

Author:

Cameron Churchwell, Max Morrison, Bryan Pardo

Keyword:

Electrical Engineering and Systems Science, Audio and Speech Processing, Audio and Speech Processing (eess.AS), Sound (cs.SD)

journal:

date:

2024-02-27 00:00:00

Abstract

A phonetic posteriorgram (PPG) is a time-varying categorical distribution over acoustic units of speech (e.g., phonemes). PPGs are a popular representation in speech generation due to their ability to disentangle pronunciation features from speaker identity, allowing accurate reconstruction of pronunciation (e.g., voice conversion) and coarse-grained pronunciation editing (e.g., foreign accent conversion). In this paper, we demonstrably improve the quality of PPGs to produce a state-of-the-art interpretable PPG representation. We train an off-the-shelf speech synthesizer using our PPG representation and show that high-quality PPGs yield independent control over pitch and pronunciation. We further demonstrate novel uses of PPGs, such as an acoustic pronunciation distance and fine-grained pronunciation control.

PDF: High-Fidelity Neural Phonetic Posteriorgrams.pdf