Audiobox by Meta

What Audiobox is

Audiobox is an AI audio generation platform developed by Meta's FAIR team. The public description covers voice cloning, text-to-speech, sound effect generation, voice style reshaping and audio completion, using self-supervised learning over large volumes of speech, music and sound effect data.

It is presented as a research and exploration platform rather than a commercial production suite.

What you can do with it

• Generate speech from a recording or a prompt
• Create sound effects
• Reshape a voice into different styles
• Complete or replace parts of an audio clip

Who it is for

• Content creators experimenting with audio
• Developers and researchers
• Game, education and brand audio teams

What to watch out for

• Voice cloning must use your own voice or licensed material; cloning others for imitation or deception is not acceptable
• Research platforms change, pause or shut down; do not build a critical workflow on one
• Generated audio rights and usage terms should be confirmed before commercial release
• Synthetic voices should be disclosed in public content
• Output quality varies by language and style