New day, new update on generative AI. In these hours Meta has in fact announced its latest AI model: it’s called Voicebox and was designed for the creation of sound clips, a support tool for visually impaired people and – why not – for content creators.
Meta announces Voicebox: what the generative AI model is for
While still in the process experimentalthe company led by Mark Zuckerberg (which recently also presented the new generation of the Meta Quest) has released a press release useful for understanding the functioning and the potential utility of Voicebox.
Meta explains that its generative AI model can produce high-quality audio clips and edit pre-recorded audio. It can remove noise from your recordings, such as a car horn or a dog barking in the background. But always preserving the content and quality of the same.
According to the US giant, multipurpose AI models like Voicebox could provide a “natural” voice to virtual assistants or give one to NPCs (non-player characters) in the metaverse. They could also allow the visually impaired to listen to written messages from friends, family and work colleagues read by the AI in a voice very similar to theirs. As for the creators, however, Meta says that Voicebox could provide them with tools for creating and editing audio tracks for videosfor example.
Responsible use: Meta speaks
According to Zuckerberg and associates, Voicebox can usher in a new era for generative AI related to speech. However, like other important innovations in the field of artificial intelligence, this model could also be misused and create some difficulties. Recognizing this danger, Meta has drafted a document complete with all the details regarding an ad-hoc classifier that is able to notice the difference between authentic speech and audio generated with Voicebox.
The use of artificial intelligence, concludes Meta, should always be responsible. That’s why, although Voicebox is still in an experimental phase, the company has already wanted to share its first results in the field of generative AI.