HiFi-GAN: High-Fidelity Denoising and Dereverberation Based on Speech Deep Features in Adversarial Networks
This addresses audio quality issues for applications like speech processing and communication, but it is incremental as it builds on existing adversarial and WaveNet architectures.
The paper tackled the problem of enhancing real-world speech recordings degraded by noise, reverberation, and distortion, introducing HiFi-GAN to transform them to studio-quality sound, achieving significant outperformance over state-of-the-art methods in objective and subjective experiments.
Real-world audio recordings are often degraded by factors such as noise, reverberation, and equalization distortion. This paper introduces HiFi-GAN, a deep learning method to transform recorded speech to sound as though it had been recorded in a studio. We use an end-to-end feed-forward WaveNet architecture, trained with multi-scale adversarial discriminators in both the time domain and the time-frequency domain. It relies on the deep feature matching losses of the discriminators to improve the perceptual quality of enhanced speech. The proposed model generalizes well to new speakers, new speech content, and new environments. It significantly outperforms state-of-the-art baseline methods in both objective and subjective experiments.