
Meta announces the release of their latest AI tool that generates high-quality, realistic music and audio from text. This new open-source AI tool called AudioCraft consists of three generative AI models – AudioGen, EnCodec, and MusicGen.MusicGen is a tool that generates music from text prompts. It was trained using over 20,000 hours of music that is Meta-owned or licensed for this purpose. Meta’s EnCodec allows higher-quality music generation with fewer artifacts preventing audio manipulation from causing distortion.AudioGen generates audio from text prompts and it was trained on public sound effects.
Meta says that they’ve seen a lot of excitement around generative AI for images, video, and text, but, it lacks behind in terms of audio. This forces Meta to take this new move to provide a platform that generates audio easily. The AudioCraft produces high-quality audio with long-term consistency. The company said that AudioCraft works for music, sound, compression, and generation- all in the same place. Ease of use and reusability are spotted as the advantages of this AudioCraft.Individuals who have a keen interest in developing better sound generators, compression algorithms, or music generators can make use of this AI generative tool.
It requires technical skills to use the tool effectively as AudioCraft is not intended for regular users. The company states that it is primarily designed for research purposes. The developers are trying to enhance the performance and control methods of these models to expand their capabilities. The company has provided audio samples on its website for people to check the ability of the new AI.
- Can Your Company Blog Be Featured In Google News? - December 20, 2023
- Bringing Back Vision: Smart Glasses Help the Blind “See” Through Sound - November 2, 2023
- Apple intends to largely push generative AI into its devices and services - October 25, 2023