SpecGrad: Diffusion Probabilistic Model based Neural Vocoder with Adaptive Noise Spectral Shaping

TL;DR

SpecGrad introduces an adaptive noise spectral shaping method for diffusion-based neural vocoders, significantly enhancing high-frequency sound quality without increasing computational costs.

Contribution

It proposes a novel adaptive noise spectral shaping technique that aligns the diffusion noise with the spectral envelope of acoustic features in neural vocoders.

Findings

Higher-fidelity speech synthesis compared to conventional DDPM vocoders

Effective in analysis-synthesis and speech enhancement scenarios

Maintains computational efficiency similar to existing models

Abstract

Neural vocoder using denoising diffusion probabilistic model (DDPM) has been improved by adaptation of the diffusion noise distribution to given acoustic features. In this study, we propose SpecGrad that adapts the diffusion noise so that its time-varying spectral envelope becomes close to the conditioning log-mel spectrogram. This adaptation by time-varying filtering improves the sound quality especially in the high-frequency bands. It is processed in the time-frequency domain to keep the computational cost almost the same as the conventional DDPM-based neural vocoders. Experimental results showed that SpecGrad generates higher-fidelity speech waveform than conventional DDPM-based neural vocoders in both analysis-synthesis and speech enhancement scenarios. Audio demos are available at wavegrad.github.io/specgrad/.