FlashSR: One-step Versatile Audio Super-resolution via Diffusion Distillation

TL;DR

FlashSR is a fast, single-step diffusion model that significantly improves versatile audio super-resolution, restoring high-frequency components efficiently across various audio domains.

Contribution

The paper introduces FlashSR, a novel one-step diffusion-based approach with diffusion distillation, achieving rapid inference for high-quality 48kHz audio super-resolution.

Findings

Achieves 22x faster inference than previous diffusion models.

Maintains competitive performance in objective and subjective evaluations.

Introduces SR Vocoder for enhanced mel-spectrogram-based super-resolution.

Abstract

Versatile audio super-resolution (SR) is the challenging task of restoring high-frequency components from low-resolution audio with sampling rates between 4kHz and 32kHz in various domains such as music, speech, and sound effects. Previous diffusion-based SR methods suffer from slow inference due to the need for a large number of sampling steps. In this paper, we introduce FlashSR, a single-step diffusion model for versatile audio super-resolution aimed at producing 48kHz audio. FlashSR achieves fast inference by utilizing diffusion distillation with three objectives: distillation loss, adversarial loss, and distribution-matching distillation loss. We further enhance performance by proposing the SR Vocoder, which is specifically designed for SR models operating on mel-spectrograms. FlashSR demonstrates competitive performance with the current state-of-the-art model in both objective and subjective evaluations while being approximately 22 times faster.