Datasets Needed for Tutorials and How-to-guides

[rbracco]

We need to decide on datasets to use in the library. The primary purposes of the datasets will be

1. Benchmarking results to show efficacy of the library
2. Benchmarking results to see which transforms work in which subdomains
3. Teaching users to build their own AudioML models.

We need datasets for some of the following subdomains:

• Scene Classification
  • [Kaggle Freesound 2018 7GB] (https://www.kaggle.com/c/freesound-audio-tagging/data)
  • [ESC-50 600MB] (https://github.com/karolpiczak/ESC-50)
• Voice Recognition
  • [VoxCeleb I & II Huge] (http://www.robots.ox.ac.uk/~vgg/data/voxceleb/)
  • [Mozilla CommonVoice 38GB] (https://voice.mozilla.org/en/datasets)
• Command/Keyword Recognition
  • [Kaggle Tensorflow Command 4GB] (https://www.kaggle.com/c/tensorflow-speech-recognition-challenge/)
  • [Free Spoken Digit Dataset (MNISTish) 10MB] (https://github.com/Jakobovski/free-spoken-digit-dataset)
• Speech Recognition
  • [Librispeech 6.3GB-60GB] (http://www.openslr.org/12/)
  • [TIMIT Dataset 440MB] (http://academictorrents.com/details/34e2b78745138186976cbc27939b1b34d18bd5b3)
  • [Mozilla CommonVoice 38GB] (https://voice.mozilla.org/en/datasets)
• Music Information Retrieval
  • [Nsynth Annotated Musical Notes 24GB] (https://magenta.tensorflow.org/datasets/nsynth)
  • [Maestro MIDI files 121GB] (https://magenta.tensorflow.org/datasets/maestro#v200)

[johanndiedrick]

@rbracco I'm working on the Kaggle Freesound one, as a way to give me an example to work on to fix the display issue