Meta has just introduced MusicGen, the company’s generative open-sourced AI that can make music based on your prompts. It does so through iterative training and fine-tuning done by the company.
Meta – MusicGen
The new generative AI will let you enter a brief description of what kind of track you want and the AI will generate a 12-second track based on this.
The current iteration of the AI only focuses on instrumental only, which should make it ideal to add a little dramatic or personal vibe in presentations or videos.
Meta uses 10,000 high-quality licensed music tracks as well as royalty-free songs from platforms like Shutterstock and Pond5.
While this is quite interesting, this could put Meta in the crosshairs of artists and labels that have begun pursuing legal action against developers and platforms for copyright infringement.
A demo of MusicGen can be tried out via Hugging Face’s API, the code can also be downloaded via GitHub but you will need around 16GB of VRAM on your GPU.
Source: Gadget Pilipinas
0 Comments