Meta, the mum or dad firm of Fb and Instagram, introduced a speech-generation AI mannequin referred to as Voicebox on June 16.
The corporate mentioned Voicebox may generate speech from textual content and famous that the mannequin may match an audio fashion primarily based on a pattern simply two seconds lengthy.
Voicebox can even convert a textual content pattern to a different language and, given a separate speech pattern, learn the translated textual content within the speaker’s authentic voice. This functionality helps six languages: English, French, German, Spanish, Polish, and Portuguese.
The AI mannequin can moreover edit present recordings to take away background noise. Extra typically, it could actually create speech that’s modeled on numerous speech samples.
Voicebox may very well be leveraged by varied customers
Meta mentioned that Voicebox and different related AI fashions may enable digital assistants and non-player characters in its metaverse to have lifelike voices. The software may be of use to content material creators and to customers with accessibility wants, it mentioned.
Meta mentioned that Voicebox is at present a analysis mission. It didn’t say when the characteristic may be publicly obtainable, but it surely shared a demo video.
Meta introduced a number of shopper AI instruments earlier in June, revealed particulars about its AI chips in Could, and mentioned inner AI purposes in an April investor name.
The put up Meta unveils speech technology AI: Voicebox appeared first on CryptoSlate.
