Facebook’s parent company Meta introduced Voicebox, an artificial intelligence tool that lets you edit audio and speak in six languages. This application can be used in the future to give virtual assistants more natural sound.
Meta writes in a blog post that the new tool can perform speech generation tasks that include text-to-speech synthesis, speech editing, noise reduction, and cross-language style transfer.
Voicebox can recreate a part of a speech that was interrupted by noise or replace misspoken words without having to re-record. The tool’s ability can create audio files in English, French, German, Spanish, Polish or Portuguese based on a sample speech.
In the future multipurpose generative artificial intelligence models like Voicebox may provide natural-sounding voices for virtual assistants. Similar solutions will allow the visually impaired to hear messages written by friends in their own voice and also give crators new tools to easily create and edit video soundtracks.
Source: Meta blog
