Common Voice is an open initiative of Mozilla with which aims to give machines a human voice, in contrast to some corporations that try to do the same but in a closed way, something that for the foundation ends up hampering innovation.
Through Common Voice, according to the official website, it is intended “build an open source voice database that anyone can use to make innovative applications for devices and the web“. To do this, volunteers can donate their voice by reading a prayer so that the machines can speak like people, something for which they recommend before reviewing what has been done previously to improve the quality of the contributions.
However, this initiative had a small drawback, and that is that until now there was only a complete set of data in English, which appeared in November 2017 including 400,000 recordings of 20,000 participants that added up to 500 hours of speech. Fortunately, according to VentureBeat, Mozilla has announced the complete data sets in three other languages: French, German and Welsh. There are currently 40 other languages in progress, including Brazilian Portuguese, Dutch, Taiwanese Chinese, Georgian, Breton, Frisian, Indonesian, Russian, Chinese Chinese, Spanish, Catalan and Danish.
The data sets that Mozilla is pushing will be of open source and may be freely used by anyone to create speech recognition technologies for both services and applications. The origin of Common Voice is similar to that of OpenStreetMap, which emerged as a response to the large number of proprietary solutions that have ended up flooding the sector.
We’ll see without Common Voice getting the impact it hopes to get. The good news is that, in addition to the giants that are usually overloaded with resources, it could help small projects that cannot meet the challenge of voice recognition and assistant technologies.