Multitask Learning for Polyphonic Piano Transcription, a Case Study
This work addresses transcription accuracy for piano music, but it appears incremental as it focuses on evaluating existing multitask approaches rather than introducing a new method.
The study tackled polyphonic piano transcription by framing it as a multitask learning problem to predict note onsets, frames, and offsets, and investigated how additional prediction targets affect performance using convolutional neural networks on the MAESTRO dataset.
Viewing polyphonic piano transcription as a multitask learning problem, where we need to simultaneously predict onsets, intermediate frames and offsets of notes, we investigate the performance impact of additional prediction targets, using a variety of suitable convolutional neural network architectures. We quantify performance differences of additional objectives on the large MAESTRO dataset.