- O project Open source recognizes over 4000 spoken languages and produces text in over 1100 languages.
- Using audio recordings of religious texts, Meta collects data to train MMS.
- MMS uses the wav2vec 2.0 model (a self-supervised speech representation learning model) to achieve significant results in speech recognition.
- Meta recognizes that the model is not perfect and may result in translation errors.
- The company emphasizes the importance of responsible development of AI technologies.
- Meta's intention by launching MMS as an open source project is to reverse the disappearance of languages.
- The model was publicly disclosed to encourage research in the community.
- The goal is to enable people to speak and learn in their native languages with the help of technology.
- In this way, it is possible to keep languages alive and promote linguistic inclusion through technology.
See also: