Real-Time Band-Grouped Vocal Denoising Using Sigmoid-Driven Ideal Ratio Masking

TL;DR

This paper introduces a real-time vocal denoising method using a sigmoid-driven ideal ratio mask with band-grouped architecture, achieving low latency and improved perceptual quality.

Contribution

It presents a novel spectral loss-trained model with a band-grouped encoder-decoder architecture for low-latency, high-quality vocal denoising in real-time applications.

Findings

Achieves less than 10 ms latency.

Improves PESQ-WB by 0.21 on stationary noise.

Enhances perceptual quality of denoised vocals.

Abstract

Real-time, deep learning-based vocal denoising has seen significant progress over the past few years, demonstrating the capability of artificial intelligence in preserving the naturalness of the voice while increasing the signal-to-noise ratio (SNR). However, many deep learning approaches have high amounts of latency and require long frames of context, making them difficult to configure for live applications. To address these challenges, we propose a sigmoid-driven ideal ratio mask trained with a spectral loss to encourage an increased SNR and maximized perceptual quality of the voice. The proposed model uses a band-grouped encoder-decoder architecture with frequency attention and achieves a total latency of less than 10,ms, with PESQ-WB improvements of 0.21 on stationary noise and 0.12 on nonstationary noise.