Resource-constrained stereo singing voice cancellation
This addresses the problem of real-time singing voice processing for applications with limited memory and compute, though it is incremental as it builds on an existing mono model.
The paper tackles stereo singing voice cancellation by adapting an efficient mono speech separation model to handle stereo input, achieving performance comparable to large state-of-the-art networks while being suitable for resource-constrained real-time applications, as confirmed by objective metrics and large-scale MUSHRA trials.
We study the problem of stereo singing voice cancellation, a subtask of music source separation, whose goal is to estimate an instrumental background from a stereo mix. We explore how to achieve performance similar to large state-of-the-art source separation networks starting from a small, efficient model for real-time speech separation. Such a model is useful when memory and compute are limited and singing voice processing has to run with limited look-ahead. In practice, this is realised by adapting an existing mono model to handle stereo input. Improvements in quality are obtained by tuning model parameters and expanding the training set. Moreover, we highlight the benefits a stereo model brings by introducing a new metric which detects attenuation inconsistencies between channels. Our approach is evaluated using objective offline metrics and a large-scale MUSHRA trial, confirming the effectiveness of our techniques in stringent listening tests.