Skip to content

Dataset · facebook

Multilingual European Parliament Speech Corpus

facebook/voxpopuli

European Parliament speech audio and transcripts in many languages for ASR and representation learning.

What's inside

A large multilingual speech corpus compiled from 2009-2020 European Parliament event recordings. Contains transcribed speech for 18 languages, plus 29 hours of non-native transcripts. Modalities are audio and text; metadata indicates size 1M-10M.

What you can do with it

  • Train ASR model on European languages
  • Fine-tune multilingual speech encoder
  • Evaluate speech transcription quality
Tasks
Automatic speech recognition
Languages
English, German, French, Spanish, Polish, Italian, Romanian, Hungarian and 4 more
Size
1M–10M rows
Downloads this month
39.5K
Likes
183

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("facebook/voxpopuli")