A Flow-Based Neural Network for Time Domain Speech Enhancement
This work addresses speech enhancement for audio processing applications, presenting an incremental improvement by adapting an existing flow-based method to a new task.
The paper tackled speech enhancement by proposing a normalizing flow framework to directly model clean speech from noisy inputs in the time domain, achieving results comparable to state-of-the-art GAN-based approaches and surpassing baseline methods in objective metrics.
Speech enhancement involves the distinction of a target speech signal from an intrusive background. Although generative approaches using Variational Autoencoders or Generative Adversarial Networks (GANs) have increasingly been used in recent years, normalizing flow (NF) based systems are still scarse, despite their success in related fields. Thus, in this paper we propose a NF framework to directly model the enhancement process by density estimation of clean speech utterances conditioned on their noisy counterpart. The WaveGlow model from speech synthesis is adapted to enable direct enhancement of noisy utterances in time domain. In addition, we demonstrate that nonlinear input companding benefits the model performance by equalizing the distribution of input samples. Experimental evaluation on a publicly available dataset shows comparable results to current state-of-the-art GAN-based approaches, while surpassing the chosen baselines using objective evaluation metrics.