Meta-learning Extractors for Music Source Separation
This work addresses efficient and effective source separation for music processing, though it appears incremental as it builds on existing meta-learning and separation techniques.
The paper tackles music source separation by proposing Meta-TasNet, a hierarchical meta-learning model that uses a generator to predict weights for instrument-specific extractors, achieving performance comparable to state-of-the-art methods with fewer parameters and faster run-time.
We propose a hierarchical meta-learning-inspired model for music source separation (Meta-TasNet) in which a generator model is used to predict the weights of individual extractor models. This enables efficient parameter-sharing, while still allowing for instrument-specific parameterization. Meta-TasNet is shown to be more effective than the models trained independently or in a multi-task setting, and achieve performance comparable with state-of-the-art methods. In comparison to the latter, our extractors contain fewer parameters and have faster run-time performance. We discuss important architectural considerations, and explore the costs and benefits of this approach.