Neural Network Alternatives to Convolutive Audio Models for Source Separation
This work addresses audio source separation for speech processing applications, presenting an incremental improvement over existing methods.
The paper tackled the problem of audio source separation by proposing a convolutional auto-encoder as a neural network alternative to convolutive Non-Negative Matrix Factorization, achieving significant improvement in separation performance on speech mixtures from the TIMIT dataset as measured by BSSeval metrics.
Convolutive Non-Negative Matrix Factorization model factorizes a given audio spectrogram using frequency templates with a temporal dimension. In this paper, we present a convolutional auto-encoder model that acts as a neural network alternative to convolutive NMF. Using the modeling flexibility granted by neural networks, we also explore the idea of using a Recurrent Neural Network in the encoder. Experimental results on speech mixtures from TIMIT dataset indicate that the convolutive architecture provides a significant improvement in separation performance in terms of BSSeval metrics.