We provide a PyTorch implementation of the paper: Real Time Speech Enhancement in the Waveform Domain. In which, we present a causal speech enhancement model working on the raw waveform that runs in real-time on a laptop CPU. The proposed model is based on an encoder-decoder architecture with skip-connections. It is optimized on both time and frequency domains, using multiple loss functions. Empirical evidence shows that it is capable of removing various kinds of background noise including stationary and non-stationary noises, as well as room reverb. Additionally, we suggest a set of data augmentation techniques applied directly on the raw waveform which further improve model performance and its generalization abilities.
At the moment, we do not provide official support for other OSes. However, if you have a a soundcard that supports loopback (for instance Steinberg products), you can try to make it work. You can list the available audio interfaces with python -m sounddevice. Then once you have spotted your loopback interface, just run
denoiser can introduce distortions for very high level of noises. Audio can become crunchy if your computer is not fast enough to process audio in real time. In that case, you will see an error message in your terminal warning you that denoiser is not processing audio fast enough. You can try exiting all non required applications.