AudioLoader 0.1.2

A collection of PyTorch audio datasets for speech and music applications

AudioLoader

This will be a collection of PyTorch audio dataset loaders that are not available in the official PyTorch dataset and torchaudio dataset yet. I am building various one-click-ready audio datasets for my research, and I hope it will also benefit other people.

Currently supported datasets:

1. Speech
   1. Multilingual LibriSpeech (MLS)
   2. TIMIT
   3. SpeechCommands v2 (12 classes)
2. Automatic Music Transcription (AMT)
   1. MAPS
   2. MusicNet
   3. MAESTRO
3. Music Source Separation (MSS)
   1. FastMUSDB
   2. MusdbHQ

Installation

pip install git+https://github.com/KinWaiCheuk/AudioLoader.git

News & Changelog

version 0.0.3 (10 Sep 2021):

1. Replace broken links with a working links for MAPS and TIMIT
2. Remove the slience indicators in the phonemic labels for TIMIT