Modeling Musical Onset Probabilities via Neural Distribution Learning
This work addresses onset detection in music, an incremental improvement for audio processing applications.
The paper tackled musical onset detection by modeling onset probabilities as a time-to-event or time-since-event prediction task, achieving results comparable to previous deep-learning models on the Bock dataset.
Musical onset detection can be formulated as a time-to-event (TTE) or time-since-event (TSE) prediction task by defining music as a sequence of onset events. Here we propose a novel method to model the probability of onsets by introducing a sequential density prediction model. The proposed model estimates TTE & TSE distributions from mel-spectrograms using convolutional neural networks (CNNs) as a density predictor. We evaluate our model on the Bock dataset show-ing comparable results to previous deep-learning models.