At the end of last week, the artificial intelligence Voicebox was unveiled, which appears to be a powerful AI for generating speech from text.
Voicebox can generate different types of voices and speech based on the text it is fed. The AI can also remove background noise from recorded audio if desired. It is also possible to input a recorded voice of someone and have Voicebox read it aloud to sound like the person whose voice was sampled.
According to Meta, only two seconds of voice sampling from someone is enough to make Voicebox sound exactly like that person. Due to the potential for misuse, Meta will not release Voicebox or the AI’s code publicly for the time being. In a post, Meta states:
There are many exciting use cases for generative speech models, but because of the potential risks of misuse, we are not making the Voicebox model or code publicly available at this time. While we believe it is important to be open with the AI community and to share our research to advance the state of the art in AI, it’s also necessary to strike the right balance between openness with responsibility.
