DiffProsody: Diffusion-based Latent Prosody Generation for Expressive Speech Synthesis with Prosody Conditional Adversarial Training

TL;DR

DiffProsody introduces a diffusion-based latent prosody generator combined with adversarial training, enabling faster and more expressive speech synthesis with improved prosody quality.

Contribution

It presents a novel diffusion-based approach for prosody generation in speech synthesis, significantly enhancing speed and quality over traditional autoregressive methods.

Findings

Prosody generator effectively produces realistic prosody vectors.

Prosody conditional discriminator improves speech quality.

DiffProsody is 16 times faster than traditional diffusion models.

Abstract

Expressive text-to-speech systems have undergone significant advancements owing to prosody modeling, but conventional methods can still be improved. Traditional approaches have relied on the autoregressive method to predict the quantized prosody vector; however, it suffers from the issues of long-term dependency and slow inference. This study proposes a novel approach called DiffProsody in which expressive speech is synthesized using a diffusion-based latent prosody generator and prosody conditional adversarial training. Our findings confirm the effectiveness of our prosody generator in generating a prosody vector. Furthermore, our prosody conditional discriminator significantly improves the quality of the generated speech by accurately emulating prosody. We use denoising diffusion generative adversarial networks to improve the prosody generation speed. Consequently, DiffProsody is capable of generating prosody 16 times faster than the conventional diffusion model. The superior performance of our proposed method has been demonstrated via experiments.