Dual-Signal Transformation LSTM Network for Real-Time Noise Suppression
This work addresses noise suppression in speech for applications like communication systems, though it is incremental as it builds on existing LSTM and transformation methods.
The paper tackles real-time speech enhancement by introducing a dual-signal transformation LSTM network (DTLN) that combines STFT and learned bases, achieving state-of-the-art performance with a 0.24-point absolute improvement in mean opinion score over the DNS-Challenge baseline.
This paper introduces a dual-signal transformation LSTM network (DTLN) for real-time speech enhancement as part of the Deep Noise Suppression Challenge (DNS-Challenge). This approach combines a short-time Fourier transform (STFT) and a learned analysis and synthesis basis in a stacked-network approach with less than one million parameters. The model was trained on 500 h of noisy speech provided by the challenge organizers. The network is capable of real-time processing (one frame in, one frame out) and reaches competitive results. Combining these two types of signal transformations enables the DTLN to robustly extract information from magnitude spectra and incorporate phase information from the learned feature basis. The method shows state-of-the-art performance and outperforms the DNS-Challenge baseline by 0.24 points absolute in terms of the mean opinion score (MOS).