Meta, the parent company of Facebook and Instagram, has announced a voice-generating AI model called Voicebox. June 16th.
The company said Voicebox can generate speech from text, and said the model can match voice styles based on samples that are only two seconds long.
Voicebox can also convert a text sample into another language, and given another audio sample, it can read the translated text in the speaker’s original voice. This feature supports her six languages: English, French, German, Spanish, Polish, and Portuguese.
AI models can further edit existing recordings to remove background noise. More generally, you can create modeled speech based on different speech samples.
Voicebox is for different users
Meta said Voicebox and other similar AI models could potentially give realistic voices to virtual assistants and non-player characters within the Metaverse. The tool could also be useful for content creators and users with accessibility needs, the company said.
Meta said Voicebox is currently a research project. No word yet on when the feature will roll out to the general public, but a demo video is available.
Meta announced several consumer AI tools in early June, unveiled details of its AI chips in May, and discussed internal AI applications on an investor conference call in April.
The article Meta announces voice-generating AI: Voicebox first appeared on CryptoSlate.
