AI model ‘speaks’ 1000+ languages based on limited training data

MIT Technology Review: Meta’s new AI models can recognize and produce speech for more than 1,000 languages. The generally more interesting point, to me, was that they were able to train the model on “two new data sets: one that contains audio recordings of the New Testament Bible and its corresponding text taken from the internet in 1,107 languages, and another containing unlabeled New Testament audio recordings in 3,809 languages”.

Open source models that can perform well using limited amounts of (internal) data seems to be a good fit for enterprise use.