Byte Pair Encoding for Symbolic Music - ESEO
Communication Dans Un Congrès Année : 2023

Byte Pair Encoding for Symbolic Music

Résumé

When used with deep learning, the symbolic music modality is often coupled with language model architectures. To do so, the music needs to be tokenized, i.e. converted into a sequence of discrete tokens. This can be achieved by different approaches, as music can be composed of simultaneous tracks, of simultaneous notes with several attributes. Until now, the proposed tokenizations rely on small vocabularies of tokens describing the note attributes and time events, resulting in fairly long token sequences, and a sub-optimal use of the embedding space of language models. Recent research has put efforts on reducing the overall sequence length by merging embeddings or combining tokens. In this paper, we show that Byte Pair Encoding, a compression technique widely used for natural language, significantly decreases the sequence length while increasing the vocabulary size. By doing so, we leverage the embedding capabilities of such models with more expressive tokens, resulting in both better results and faster inference in generation and classification tasks. The source code is shared on Github, along with a companion website. Finally, BPE is directly implemented in MidiTok, allowing the reader to easily benefit from this method.
Fichier principal
Vignette du fichier
EMNLP_2023.pdf (18.37 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
licence

Dates et versions

hal-03976252 , version 1 (06-02-2023)
hal-03976252 , version 2 (09-10-2023)

Licence

Identifiants

Citer

Nathan Fradet, Nicolas Gutowski, Fabien Chhel, Jean-Pierre Briot. Byte Pair Encoding for Symbolic Music. The 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP 2023), Association for Computational Linguistics, Dec 2023, Singapore, Singapore. pp.2001-2020, ⟨10.18653/v1/2023.emnlp-main.123⟩. ⟨hal-03976252v2⟩
435 Consultations
46 Téléchargements

Altmetric

Partager

More