Audiobox – a free and open source AI speech and sound generation model launched by Meta | AI toolset

Audiobox is a free and open source AI speech and sound generation model launched by Meta on November 30, 2023. The online web version will be launched on December 11, and users can experience the capabilities of the model for free. Audiobox is Meta’s latest generation audio generation model after Voicebox, which can use a combination of voice input and natural language text prompts to generate speech and sound effects, making it easy to create realistic custom audio for a variety of use cases.

Main functions of Audiobox

1. Clone a user’s voice: record a voice in the style of the user’s voice or generate speech in the style of any audio sample
2. Text description to generate vocals: Use text to describe the characteristics of the voice style and the acoustic environment to generate the vocals
3. Change sound style: Combine sound and text descriptions to change existing sound styles
4. This article describes generating sound effects: generating sound effects based on input text descriptions of sound features
5. Noise Cancellation: Provides Magic Eraser function to eliminate transient noise in recordings
6. Sound Filling: Replace a portion of the audio with a new sound based on a textual description
7. Audio Story Maker: Combine the above functions and use Audiobox Maker to create original and interesting audio stories