Do sound event representations generalize to other audio tasks? A case study in audio transfer learning
This work addresses the problem of efficient transfer learning for audio tasks, but it is incremental as it applies an existing method to new data.
The study investigated whether audio representations from neural networks trained on a large-scale sound event detection dataset can generalize to other audio tasks using linear classifier transfer, showing that this simple approach achieves high performance.
Transfer learning is critical for efficient information transfer across multiple related learning problems. A simple, yet effective transfer learning approach utilizes deep neural networks trained on a large-scale task for feature extraction. Such representations are then used to learn related downstream tasks. In this paper, we investigate transfer learning capacity of audio representations obtained from neural networks trained on a large-scale sound event detection dataset. We build and evaluate these representations across a wide range of other audio tasks, via a simple linear classifier transfer mechanism. We show that such simple linear transfer is already powerful enough to achieve high performance on the downstream tasks. We also provide insights into the attributes of sound event representations that enable such efficient information transfer.