Meta, the parent company of Facebook and Instagram, has recently unveiled its use of the Bible and other religious texts in training its advanced multilingual Artificial Intelligence technology.
This cutting-edge AI speech model can identify over 4,000 spoken languages, transcribing and translating them simultaneously.
Previously limited to accommodate a maximum of 100 languages, Meta turned to the Bible and other religious texts due to their extensive translations into numerous languages and their extensive use as valuable resources for text-based language translation research.
This strategic decision has significantly expanded the model's language coverage, enabling it to surpass its previous limitations.
“These translations have publicly available audio recordings of people reading these texts in different languages. As part of the MMS project, we created a dataset of readings of the New Testament in more than 1,100 languages, which provided on average 32 hours of data per language,” Meta’s statement read.
By considering the recordings of various other Christian readings, the company says it increased the number of languages available from 100 to more than 4,000.
While the technology is new, Meta shares its data and code as open source. This allows others to contribute, develop, and improve the platform.