It is now possible to produce music just by writing: AudioCraft is Available!
Meta, in partnership with Microsoft, unveiled Audio Craft, its latest AI tool that can generate high-quality audio and music from text.
The tool aims to simplify the process of adding music to videos on Instagram and eliminate the need to browse multiple songs.
While some artists are apprehensive about the technology, Audio Craft is currently available as open source code to researchers and practitioners to advance the field of AI-generated sound and music.
AudioCraft consists of three models:
- An improved version of EnCodec
How Meta’s AudioCraft Artificial Intelligence Works
While MusicGen is designed to create music, it is trained on a large dataset of 400,000 music recordings.
AudioGen generates realistic environmental sounds based on written descriptions of acoustic scenes.
EnCodec produces higher quality music with fewer problems.
Recognising the importance of responsible innovation, Meta aims to address bias and abuse in generative models by sharing AudioCraft’s code and encouraging diversity in the training datasets for these models.
The company is excited to witness the creative results people will produce using AudioCraft. What are your thoughts on the future of this technology? Please share your thoughts with us in the comments section below.