Giter VIP home page Giter VIP logo

beast's Introduction

BEAST

An online beat tracking system based on streaming Transformer.

arXiv

The source code of the paper BEAST: Online Joint Beat and Downbeat Tracking Based on Streaming Transformer, accepted by ICASSP 2024.

More information to be updated.

Usage

  1. Data preparation

If you want to directly evaluate the model, you can download the preprocessed GTZAN data here and the pretrained model here. And put the model under the ./data directory.

  1. Evaluation

Run ./code/eval.py.

Contact

Acknowlegement

  1. We borrowed the code from ESPnet for contextual block processing transformer based modeling.
  2. We borrowed the code from Beat-Transformer for beat tracking.

beast's People

Contributors

wildhoneypie avatar

Stargazers

avatar JS avatar avatar avatar Chih-Wei Wu avatar MelonJack avatar Ubaid Seth avatar avatar SpectraSynq avatar aaronchen avatar eagle avatar Amantur Amatov avatar Hans Brouwer avatar jiyun.park avatar Huan Zhang avatar OOOOQII avatar SeungHeon Doh avatar yearnyeen ho avatar Sebastian Murgul avatar Mojtaba (Moji) Heydari avatar Hao Hao Tan avatar Minsu Kang avatar Sangeon Yong avatar avatar

Watchers

Kostas Georgiou avatar avatar

Forkers

ma5onic

beast's Issues

BeatNet and 1D models latencies

Hi there,

Great work! I just came across your paper and enjoyed reading it a lot. However, I noticed an error that I thought it may be worth mentioning. According to the paper, the latencies in the evaluation table are defined as the hop sizes of the other models. Therefore, the latencies for the BeatNet and 1D state space should be 20 ms as suggested in the papers and the implementations. However, they are reported as 46 ms in the evaluation.

For instance, the BeatNet model's hop size is defined on line #62 of the BeatNet inference main script.

Just wanted to bring this error to your attention. Thank you for your valuable work.

Cheers,
Moji

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.