Fact · Speech recognition
18languages
The VoxPopuli dataset card says it contains transcribed speech for 18 languages, and 29 hours of transcribed non-native English for research on accented speech.
Checked against its source on October 17, 2026
In the source's own words
“This implementation contains transcribed speech data for 18 languages. It also contains 29 hours of transcribed speech data of non-native English intended for research in ASR for accented speech”
VoxPopuli dataset card (facebook/voxpopuli), Meta AI, Hugging Face, n.d., Dataset description.
Open the original- What it measures
- VoxPopuli is built from European Parliament event recordings. The card's license tag is cc0-1.0, which permits reuse without conditions.
- What it does not show
- The speech is parliamentary, read or spoken at formal events, so it is not a model of everyday dictation. The 18 languages are the transcribed subset, not the full set of languages in the source recordings.