AudioCraft, an AI tool that creates professional music from text

According to Meta, AudioCraft integrates its three AI platforms, including MusicGen, AudioGen and EnCodec, providing the ability to create music, audio and audio compression.

MusicGen is an AI capable of creating music from input text developed by Meta since 2022. Con AudioGen specializes in creating sounds and sound effects from text, such as simulating dogs barking, car horns, steps. feet on wooden floor.

EnCodec, introduced in October 2022, is an AI-based audio codec that supports compression and decompression of audio files without loss of quality. Meta used a database of over 20,000 hours of music to train EnCodec. This data source is owned by the company or is specifically licensed for AI training purposes.

MusicGen, AudioGen, and EnCodec are bundled into AudioCraft, making this AI application high-quality sound, lasting consistency, and easy to use.

Up to now, generator AI tools (the type of AI used to generate new content after training on existing data) have been able to create images and videos from text with very high fidelity. Meanwhile, the sound field due to higher difficulty is lagging behind.

However, these tools after their launch were controversial. Many people think that the use of AI tools by artists is a violation of copyright, while others think that this helps reduce time for work.

Samuel Daniel

Update 03 August 2023