Shift-Invariant Kernel Additive Modelling for Audio Source Separation
This work improves audio source separation for applications like music processing or speech enhancement by enhancing a non-data-driven method, though it is incremental as it builds on existing KAM approaches.
The paper tackled the problem of audio source separation by addressing the limitation of Kernel Additive Modelling (KAM) in assuming source repetition in time and frequency, which does not always hold in practice. They introduced a shift-invariant kernel function to identify similar spectral content under frequency shifts, increasing suitable sound material and separation performance, with acceleration techniques to reduce computational complexity.
A major goal in blind source separation to identify and separate sources is to model their inherent characteristics. While most state-of-the-art approaches are supervised methods trained on large datasets, interest in non-data-driven approaches such as Kernel Additive Modelling (KAM) remains high due to their interpretability and adaptability. KAM performs the separation of a given source applying robust statistics on the time-frequency bins selected by a source-specific kernel function, commonly the K-NN function. This choice assumes that the source of interest repeats in both time and frequency. In practice, this assumption does not always hold. Therefore, we introduce a shift-invariant kernel function capable of identifying similar spectral content even under frequency shifts. This way, we can considerably increase the amount of suitable sound material available to the robust statistics. While this leads to an increase in separation performance, a basic formulation, however, is computationally expensive. Therefore, we additionally present acceleration techniques that lower the overall computational complexity.