Efficient Monaural Speech Enhancement using Spectrum Attention Fusion
This work addresses efficiency issues in speech enhancement for automated processing systems, but it is incremental as it builds on existing Transformer methods.
The authors tackled the problem of high computational cost and data requirements in Transformer-based speech enhancement by introducing Spectrum Attention Fusion, which achieved comparable or better results than state-of-the-art models with only 0.58M parameters on the Voice Bank + DEMAND dataset.
Speech enhancement is a demanding task in automated speech processing pipelines, focusing on separating clean speech from noisy channels. Transformer based models have recently bested RNN and CNN models in speech enhancement, however at the same time they are much more computationally expensive and require much more high quality training data, which is always hard to come by. In this paper, we present an improvement for speech enhancement models that maintains the expressiveness of self-attention while significantly reducing model complexity, which we have termed Spectrum Attention Fusion. We carefully construct a convolutional module to replace several self-attention layers in a speech Transformer, allowing the model to more efficiently fuse spectral features. Our proposed model is able to achieve comparable or better results against SOTA models but with significantly smaller parameters (0.58M) on the Voice Bank + DEMAND dataset.