English

A Transformer Based Pitch Sequence Autoencoder with MIDI Augmentation

Sound 2021-02-02 v3 Machine Learning Audio and Speech Processing

Abstract

Despite recent achievements of deep learning automatic music generation algorithms, few approaches have been proposed to evaluate whether a single-track music excerpt is composed by automatons or Homo sapiens. To tackle this problem, we apply a masked language model based on ALBERT for composers classification. The aim is to obtain a model that can suggest the probability a MIDI clip might be composed condition on the auto-generation hypothesis, and which is trained with only AI-composed single-track MIDI. In this paper, the amount of parameters is reduced, two methods on data augmentation are proposed as well as a refined loss function to prevent overfitting. The experiment results show our model ranks 3rd3^{rd} in all the 77 teams in the data challenge in CSMT(2020). Furthermore, this inspiring method could be spread to other music information retrieval tasks that are based on a small dataset.

Keywords

Cite

@article{arxiv.2010.07758,
  title  = {A Transformer Based Pitch Sequence Autoencoder with MIDI Augmentation},
  author = {Mingshuo Ding and Yinghao Ma},
  journal= {arXiv preprint arXiv:2010.07758},
  year   = {2021}
}

Comments

submitted to CSMT 2020 challenge track

R2 v1 2026-06-23T19:22:34.525Z