Systematic Analysis of Music Representations from BERT

June 06, 2023 ยท Entered Twilight ยท ๐Ÿ› arXiv.org

๐Ÿ’ค TWILIGHT: Eternal Rest
Repo abandoned since publication

Repo contents: .gitignore, BERT.py, README.md, config.json, convert_remi.py, data.py, requirements.txt, test.ipynb, train.py, utils

Authors Sangjun Han, Hyeongrae Ihm, Woohyung Lim arXiv ID 2306.04628 Category cs.SD: Sound Cross-listed cs.MM, eess.AS Citations 2 Venue arXiv.org Repository https://github.com/sjhan91/MusicBERT โญ 27 Last Checked 3 months ago
Abstract
There have been numerous attempts to represent raw data as numerical vectors that effectively capture semantic and contextual information. However, in the field of symbolic music, previous works have attempted to validate their music embeddings by observing the performance improvement of various fine-tuning tasks. In this work, we directly analyze embeddings from BERT and BERT with contrastive learning trained on bar-level MIDI, inspecting their musical information that can be obtained from MIDI events. We observe that the embeddings exhibit distinct characteristics of information depending on the contrastive objectives and the choice of layers. Our code is available at https://github.com/sjhan91/MusicBERT.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Sound