2023-06-13 22:34:35
goal has presented MusicGena model of artificial intelligence (AI) that can generate high-quality music from a short text description and a melody.
MusicGen is a simple language model, developed under a single-stage approach with efficient interleaved token patterns -with 32Khz audio EnCodec-, which reduces the need to cascade multiple models, as researchers from Audiocraft.
From this approach, MusicGen generates high-quality music samples, about twelve seconds long, just by entering a brief text description and a reference to the melody you want to obtain.
The technology company has published on the Hugging Face AI portal a demonstration of MusicGen, with which any user can generate a sample of music. It also adds a series of examples of the descriptions, in some cases accompanied by a melodic description.
The researchers have used more than 20,000 hours of licensed music for training, including 10,000 high-quality tracks from an internal database, and data taken from Shutterstock y Pond5.
#Meta #introduces #tool #generate #highquality #music #texts #melodies