More free data for French

[Alexander Gorodetski]

Hello guys,

I am using following free data to train French model (~1200 hours).

African

LibriSpeech

TedX

CommonVoice

Can someone recommend more free data? I am especially interested in narrow band data.

Is it possible to download Netflix movies with subtitles for French only. Can someone recommend the relevant tool for such downloading process?

Thanks,

AlexG.

[entn-at]

Here are some datasets (disclaimer: I haven't used them myself, so I can't tell how useful they are):

VoxPopuli (https://github.com/facebookresearch/voxpopuli) apparently contains 211 hours of transcribed French speech (according to the paper).

MPF (Multicultural Paris French, ~19 hours): https://www.ortolang.fr/market/corpora/mpf/v3

CFPP2000 (Corpus de Français Parlé Parisien, ~16 hours): https://www.ortolang.fr/market/corpora/cfpp2000

TCOF (Traitement de Corpus Oraux en Français, apparently ~146 hours, mixed adult and child): https://www.ortolang.fr/market/corpora/tcof/v2.1

CLAPI (Corpus de LAngues Parlées en Interaction): https://www.ortolang.fr/market/corpora/clapi

Europarl-ST (https://www.mllp.upv.es/europarl-st/) also contains French speech, but might overlap with VoxPopuli.