GLA-Grad: A Griffin-Lim Extended Waveform Generation Diffusion Model
This addresses a specific challenge in speech generation for unseen speakers, representing an incremental improvement over existing diffusion models.
The paper tackled the problem of diffusion models struggling with noise diffusion and generating high-quality speech for unseen speakers by proposing GLA-Grad, which integrates the Griffin-Lim algorithm into the diffusion process, and it outperformed state-of-the-art models, particularly for unseen speakers.
Diffusion models are receiving a growing interest for a variety of signal generation tasks such as speech or music synthesis. WaveGrad, for example, is a successful diffusion model that conditionally uses the mel spectrogram to guide a diffusion process for the generation of high-fidelity audio. However, such models face important challenges concerning the noise diffusion process for training and inference, and they have difficulty generating high-quality speech for speakers that were not seen during training. With the aim of minimizing the conditioning error and increasing the efficiency of the noise diffusion process, we propose in this paper a new scheme called GLA-Grad, which consists in introducing a phase recovery algorithm such as the Griffin-Lim algorithm (GLA) at each step of the regular diffusion process. Furthermore, it can be directly applied to an already-trained waveform generation model, without additional training or fine-tuning. We show that our algorithm outperforms state-of-the-art diffusion models for speech generation, especially when generating speech for a previously unseen target speaker.