DiffWave: A Versatile Diffusion Model for Audio Synthesis
This addresses the problem of slow and low-quality audio synthesis for applications like speech generation and music production, offering a versatile and efficient alternative.
The authors tackled audio synthesis by proposing DiffWave, a diffusion model for waveform generation, achieving speech quality matching a strong WaveNet vocoder (MOS: 4.44 vs. 4.43) with much faster synthesis and outperforming autoregressive and GAN-based models in unconditional generation.
In this work, we propose DiffWave, a versatile diffusion probabilistic model for conditional and unconditional waveform generation. The model is non-autoregressive, and converts the white noise signal into structured waveform through a Markov chain with a constant number of steps at synthesis. It is efficiently trained by optimizing a variant of variational bound on the data likelihood. DiffWave produces high-fidelity audios in different waveform generation tasks, including neural vocoding conditioned on mel spectrogram, class-conditional generation, and unconditional generation. We demonstrate that DiffWave matches a strong WaveNet vocoder in terms of speech quality (MOS: 4.44 versus 4.43), while synthesizing orders of magnitude faster. In particular, it significantly outperforms autoregressive and GAN-based waveform models in the challenging unconditional generation task in terms of audio quality and sample diversity from various automatic and human evaluations.