Preprint
Common Voice: A Massively-Multilingual Speech Corpus
Rosana Ardila et al., arXiv (LREC 2020), 2020
Read the originalRead on October 16, 2026
- What it is
- The paper describing Mozilla's crowdsourced Common Voice corpus of transcribed speech in many languages, with a transfer-learning experiment.
- How much weight it carries
- Accepted at LREC 2020 and written by the Mozilla team running the project. Its size figures are a November 2019 snapshot; the corpus has grown since.
Facts taken from this source
- By the time of its 2019 paper, Mozilla's Common Voice corpus held 2,500 hours of audio from over 50,000 contributors, across 38 languages collecting data.
“Over 50,000 individuals have participated so far, resulting in 2,500 hours of collected audio.” (Abstract)
- The Common Voice paper says the corpus is the largest audio corpus in the public domain for speech recognition, by both hours and number of languages.
“To our knowledge this is the largest audio corpus in the public domain for speech recognition, both in terms of number of hours and number of languages” (Abstract)
Pages that rest on this source
- Speech Recognition: What It Is and How Well It WorksTopic
- Speech Recognition Benchmarks and LeaderboardsGuide
- Transcribe and Translate Audio in Other LanguagesGuide
- How are speech recognition models trained?Question
- How good is Whisper at Hindi?Question
- What are the best datasets for speech recognition?Question