Impact of time and note duration tokenizations on deep learning symbolic music modeling
This work addresses tokenization challenges in symbolic music modeling for tasks like generation and classification, but it is incremental as it compares existing methods.
The paper tackled the problem of how tokenization methods for symbolic music impact deep learning model performance, finding that explicit information representation leads to better results depending on the task.
Symbolic music is widely used in various deep learning tasks, including generation, transcription, synthesis, and Music Information Retrieval (MIR). It is mostly employed with discrete models like Transformers, which require music to be tokenized, i.e., formatted into sequences of distinct elements called tokens. Tokenization can be performed in different ways. As Transformer can struggle at reasoning, but capture more easily explicit information, it is important to study how the way the information is represented for such model impact their performances. In this work, we analyze the common tokenization methods and experiment with time and note duration representations. We compare the performances of these two impactful criteria on several tasks, including composer and emotion classification, music generation, and sequence representation learning. We demonstrate that explicit information leads to better results depending on the task.