Fine-grained Noise Control for Multispeaker Speech Synthesis

TL;DR

This paper introduces an unsupervised approach to fine-grained noise and prosody modeling in speech synthesis, improving disentanglement of speech attributes and enhancing expressiveness in generated speech.

Contribution

It proposes a novel framework combining adversarial training, representation bottleneck, and FVAE for detailed noise and prosody modeling in TTS systems.

Findings

Enhanced disentanglement of noise and speech factors

More expressive and natural speech synthesis

Effective frame-level noise representation learning

Abstract

A text-to-speech (TTS) model typically factorizes speech attributes such as content, speaker and prosody into disentangled representations.Recent works aim to additionally model the acoustic conditions explicitly, in order to disentangle the primary speech factors, i.e. linguistic content, prosody and timbre from any residual factors, such as recording conditions and background noise.This paper proposes unsupervised, interpretable and fine-grained noise and prosody modeling. We incorporate adversarial training, representation bottleneck and utterance-to-frame modeling in order to learn frame-level noise representations. To the same end, we perform fine-grained prosody modeling via a Fully Hierarchical Variational AutoEncoder (FVAE) which additionally results in more expressive speech synthesis.