Common Voice: A Massively-Multilingual Speech Corpus

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4218–4222 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4218 Common Voice: A Massively-Multilingual Speech Corpus Rosana Ardila,† Megan Branson,† Kelly Davis,† Michael Henretty, Michael Kohler, Josh Meyer,◦ Reuben Morais,† Lindsay Saunders,† Francis M. Tyers,‡ Gregor Weber† † Mozilla ‡ Indiana University ◦ Artie, Inc. Various Cities Bloomington, IN, USA Los Angeles, CA, USA {rosana, mbranson, kdavis, reuben, lsaunders, gweber}@mozilla.com, ftyers@iu.edu, michael.henretty@gmail.com, me@michaelkohler.info, josh.meyer@artie.com Abstract The Common Voice corpus is a massively-multilingual collection of transcribed speech intended for speech technology research and development. Common Voice is designed for Automatic Speech Recognition purposes but can be useful in other domains (e.g. language identification). To achieve scale and sustainability, the Common Voice project employs crowdsourcing for both data collection and data validation. The most recent release includes 29 languages, and as of November 2019 there are a total of 38 languages collecting data. Over 50,000 individuals have participated so far, resulting in 2,500 hours of collected audio. 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. As an example use case for Common Voice, we present speech recognition experiments using Mozilla’s DeepSpeech Speech-to-Text toolkit. By applying transfer learning from a source English model, we find an average Character Error Rate improvement of 5.99 ± 5.48 for twelve target languages (German, French, Italian, Turkish, Catalan, Slovenian, Welsh, Irish, Breton, Tatar, Chuvash, and Kabyle). For most of these languages, these are the first ever published results on end-to-end Automatic Speech Recognition. Keywords: spoken corpus, Automatic Speech Recognition, low-resource languages 1. Introduction The Common Voice project1 is a response to the current state of affairs in speech technology, in which training data is either prohibitively expensive or unavailable for most languages (Roter, 2019). We believe that speech technol- ogy (like all technology) should be open and decentralized, and the Common Voice project achieves this goal via a mix of community building, open source tooling, and a permis- sive licensing scheme. The corpus is designed to organ- ically scale to new languages as community members use the provided tools to translate the interface, submit text sen- tences, and finally record and validate voices in their new language 2. The project was started with an initial focus on English in July 2017 and then in June 2018 was made available for any language. The remainder of the paper is organized as follows: In Sec- tion (2) we motivate Common Voice and review previous multilingual corpora. Next, in Section (3) we describe the recording and validation process used to create the corpus. Next, in Section (4) we describe the current contents of Common Voice, and lastly in Section (5) we show multi- lingual Automatic Speech Recognition experiments using the corpus. 2. Prior work Some notable multilingual speech corpora include Vox- Forge (VoxForge, 2019), Babel (Gales et al., 2014), and M- AILABS (M-AILABS, …

Content truncated to 500 words