HGCN: Harmonic gated compensation network for speech enhancement
This work addresses speech enhancement for audio processing applications, but it appears incremental as it builds on existing convolutional recurrent networks with added harmonic detection and gating mechanisms.
The paper tackles the challenge of speech enhancement when harmonics are partially masked by noise by proposing a harmonic gated compensation network (HGCN), which achieves substantial gain over advanced approaches.
Mask processing in the time-frequency (T-F) domain through the neural network has been one of the mainstreams for single-channel speech enhancement. However, it is hard for most models to handle the situation when harmonics are partially masked by noise. To tackle this challenge, we propose a harmonic gated compensation network (HGCN). We design a high-resolution harmonic integral spectrum to improve the accuracy of harmonic locations prediction. Then we add voice activity detection (VAD) and voiced region detection (VRD) to the convolutional recurrent network (CRN) to filter harmonic locations. Finally, the harmonic gating mechanism is used to guide the compensation model to adjust the coarse results from CRN to obtain the refinedly enhanced results. Our experiments show HGCN achieves substantial gain over a number of advanced approaches in the community.