https://github.com/yjlolo/music-seq2seq

https://github.com/yjlolo/music-seq2seq

Last synced: 3 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/yjlolo/music-seq2seq
• Owner: yjlolo
• Created: 2018-11-06T03:36:23.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-11-14T03:55:21.000Z (over 6 years ago)
• Last Synced: 2025-01-31T06:42:42.441Z (5 months ago)
• Language: Python
• Size: 56.6 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# music-seq2seq

This is to build a seq2seq auto-encoder for music audio, and use the learnt representations for downstream tasks.

The current downstream task in interest is emotion recognition, using the [PMEmo Dataset](http://pmemo.hellohui.cn).

The repo will not include the dataset.

Under construction

- [x] dataset

- [x] data loader

    - [x] chunks division

- [x] model

    - [x] attention

    - [x] regressive inference 

    - [ ] Hierarchical RNN

    - [ ] RCNN

- [x] trainer

    - [x] extra loss constraints

- [ ] classifier

This is also the first repo, intended to improve coding skills as well.

The structure and template are adapted from [Pytorch-template](https://github.com/victoresque/pytorch-template).