asteroid Logo
v0.3.1

Start here

  • What is Asteroid?
    • Who is it for?
  • Installation
    • CUDA and PyTorch
    • Pip
    • Development installation

Asteroid

  • What is a recipe?
    • How is it organized?
    • How does it work?
  • Datasets and tasks
    • Speech separation
      • wsj0-2mix dataset
      • WHAM dataset
      • WHAMR dataset
      • LibriMix dataset
      • Kinect-WSJ dataset
      • SMS_WSJ dataset
    • Speech enhancement
      • DNS Challenge’s dataset
    • Music source separation
      • MUSDB18 Dataset
    • Environmental sound separation
      • FUSS dataset
    • Audio-visual source separation
      • AVSpeech dataset
    • Speaker extraction
  • Training and Evaluation
    • Training with PyTorchLightning
    • Evaluation
  • Pretrained models
    • Using them
    • Model caching
    • Share your models
    • Note about licenses
  • FAQ
    • My results are worse than the ones reported in the README, why?
    • How long does it take to train a model?
    • Can I use the pretrained models for commercial purposes?

Notebooks and Tutorials

  • NB: Getting stated with Asteroid
  • NB: Understanding the Filterbank API
  • NB: Our PITLossWrapper explained
  • Community NB: Numpy vs. Asteroid STFT

Package reference

  • PyTorch Datasets
    • LibriMix
    • Wsj0mix
    • WHAM!
    • WHAMR!
    • SMS-WSJ
    • KinectWSJMix
    • DNSDataset
    • MUSDB18
    • FUSS
    • AVSpeech
  • Filterbank API
    • Filterbank, Encoder and Decoder
    • Learnable filterbanks
      • Free
      • Analytic Free
      • Parameterized Sinc
    • Fixed filterbanks
      • STFT
      • MP-GTFB
    • Transforms
      • Griffin-Lim and MISI
      • Complex transforms
  • DNN building blocks
    • Convolutional blocks
    • Recurrent blocks
    • Norms
  • Models
    • Base classes
    • Ready-to-use models
    • Publishing models
  • Losses & Metrics
    • Permutation invariant training (PIT) made easy
    • Available loss functions
      • MSE
      • SDR
      • PMSQE
      • STOI
      • MultiScale Spectral Loss
      • Deep clustering (Affinity) loss
    • Computing metrics
  • Lightning Wrapper
  • Optimizers & Schedulers
    • Optimizers
    • Schedulers
  • DSP Modules
  • Utils
    • Parser utils
    • Torch utils
    • Hub utils
    • Generic utils
  • CLI

Community

  • Asteroid High-Level Contribution Guide
    • The Asteroid Contribution Process
    • Getting Started
      • Proposing new features
      • Reporting Issues
      • Implementing Features or Fixing Bugs
      • Adding Tutorials
      • Improving Documentation & Tutorials
      • Participating in online discussions
      • Submitting pull requests to fix open issues
      • Reviewing open pull requests
      • Improving code readability
      • Adding test cases to make the codebase more robust
      • Promoting Asteroid
      • Triaging issues
    • About open source development
    • Common Mistakes To Avoid
    • Frequently asked questions
    • Attribution
  • How to contribute
    • Source code contributions
    • Writing new recipes.
    • Improving the docs.
    • Coding style
asteroid
  • Docs »
  • Search
  • Edit on GitHub

© Copyright 2019, Oncoming Revision bf60ff4b.

Built with Sphinx using a theme provided by Read the Docs.