Mozilla Company updated voice data sets Common Voice, including pronunciation examples from more than 200 thousand people. Data published as public domain (CC0). The proposed sets can be used in machine learning systems to build speech recognition and synthesis models. Compared to the previous update, the volume of speech material in the collection increased from 31.1 to 31.8 thousand hours of speech, of which 20.8 thousand hours passed the verification procedure. The number of supported languages has increased from 124 to 129 (African tribal languages have been added braid, Kalenjin, kidavida, doluo And Tswana).
93.3 thousand people took part in the preparation of materials in English, dictating 3554 hours of speech (there were 92.3 thousand participants and 3508 hours). The set for the Belarusian language covers 8400 participants and 1815 hours of speech material (there were 8291 participants and 1766 hours), Russian – 3241 participants and 277 hours (there were 3206 participants and 274 hours), Uzbek – 2189 participants and 265 hours (there were 2170 participants and 264 hours), Ukrainian language – 1091 participants and 113 hours (there were 1075 participants and 112 hours).
The Common Voice project aims to organize joint work to accumulate a database of voice patterns that takes into account the diversity of voices and speech styles. Users are invited to voice phrases displayed on the screen or evaluate the quality of data added by other users. The accumulated database with records of various pronunciations of typical phrases of human speech can be used without restrictions in machine learning systems and in research projects.
Thanks for reading: