Phase-Aware Deep Speech Enhancement: It's All About The Frame Length
This work addresses latency issues in speech enhancement for real-time applications, but it is incremental as it builds on existing phase-aware DNN approaches.
The study tackled the problem of algorithmic latency in speech enhancement by investigating the role of phase and magnitude in DNN-based methods across different frame lengths, finding that DNNs can effectively estimate phase with short frames, achieving similar or better performance than longer frames, enabling low-latency high-quality enhancement.
Algorithmic latency in speech processing is dominated by the frame length used for Fourier analysis, which in turn limits the achievable performance of magnitude-centric approaches. As previous studies suggest the importance of phase grows with decreasing frame length, this work presents a systematical study on the contribution of phase and magnitude in modern Deep Neural Network (DNN)-based speech enhancement at different frame lengths. Results indicate that DNNs can successfully estimate phase when using short frames, with similar or better overall performance compared to using longer frames. Thus, interestingly, modern phase-aware DNNs allow for low-latency speech enhancement at high quality.