Self-supervised speech unit discovery from articulatory and acoustic features using VQ-VAE.

Marc-Antoine Georges Jean-Luc Schwartz Thomas Hueber

Published in: CoRR (2022)

Keyphrases

acoustic features
speech signal
automatic speech recognition
speaker verification
vector quantization

visual features
music information retrieval
speaker recognition
speech recognition
audio stream

image compression
vocal tract
mel frequency cepstral coefficients
cross correlation
audio features

audio visual
knowledge discovery
speaker identification
noisy environments
hidden markov models