Common Voice what it measures, how it is scored and who leads it

Name: Common Voice
Creator: Mozilla
Published: 2019-12-13

Speech · Mozilla · introduced Dec 13, 2019 · Word error rate

About this benchmark

Task: Transcribe crowd-sourced read speech in many languages.
Dataset: A growing multilingual corpus recorded and validated by volunteers, released in numbered versions.
Method: Word or character error rate per language.
Metric: Word error rate
Organization: Mozilla
Introduced: Dec 13, 2019
Paper: Common Voice: A Massively-Multilingual Speech Corpus (arXiv 1912.06670)
Home: commonvoice.mozilla.org/datasets

Newest first. New versions are added, never rewritten.

Version	Date	What changed
PaperThe corpus paper; datasets are released regularly with a version number.	Dec 13, 20196 years ago	The corpus paper; datasets are released regularly with a version number.

Sources: each benchmark's paper (arXiv) and its official site or leaderboard. Scores are the boards captured here every day, exactly as published.