This repository contains open data indended for Natural Language Processing research and applications. The sharing of this data has been authorized by their respective authors, with a preference for permissive licenses whenever possible.
Should you be the owner of any of the data and wish for its removal from this dataset and/or Anaouder's training data, kindly notify me.
If you are interested in contributing to this dataset by providing your own data to improve breton open speech-to-text technologies, feel free to get in touch !
Each sub-directory contains segmented audio files, and the corresponding segment text is available in the audio_text/metadata.tsv file.
The transcription may contain special tokens, such as <C'HOARZH>, <PASAAT>, <SONEREZH>...
While this dataset isn't available on HuggingFace yet, it follows HuggingFace's audio Dataset structure. After cloning this repository locally, you can easily import the data with :
from datasets import load_dataset
roadennou = load_dataset("audiofolder", data_dir="roadennou/audio_text")
# Rename column names to match MCV's
roadennou = roadennou_ouzhpenn.rename_column("transcript", "sentence")Total audio duration : 5h 34mn 7s
| Directory | Type | Source | Author(s) | Licence | Duration |
|---|---|---|---|---|---|
| An_Drouizig | Dialog, Presentation | www.youtube.com/watch?v=A-VlNvJ3alI | Pêr Morvan, Gwenn Meynier | CC BY-NC-SA | 27mn 26s |
| Becedia_Komzoù-brezhoneg | Interviews | www.bcd.bzh/becedia/fr/catherine-quiniou-tine-plounevez-du-faou www.bcd.bzh/becedia/fr/suzanne-goarnisson-scrignac |
Lors Jouin, Francis Favereau | CC BY-NC-SA | 9mn 17s |
| Brendan-Budok_Durand-Le-Ludec | Chess Youtube channel | www.youtube.com/watch?v=5K0nWjc7XZ4 | Bredan-Budok Durand-Le Ludec, Manon Jouitteau, YF Le Gall, Mélanie Jouitteau | CC BY | 5mn 30s |
| Neus_Ket_Deus_Archoazh | Documentary | www.youtube.com/watch?v=ixAnomWd1YU | Laors Skavenneg, Korin ar Mero | CC BY | 26mn 5s |
| #Brezhoneg | Read out articles | https://www.youtube.com/channel/UCv8UTrUchQafFAl2QPc2lEA | Skol an Emzav, Dizale | All rights reserved | 4h 25mn 49s |