Facebook parent company, Meta, has debuted AudioCraft, a cutting-edge generative AI tool. By employing straightforward language cues, this technology enables people to create high-quality audio and musical content. The family of models known as AudioCraft, which includes MusicGen, AudioGen, and EnCodec, uses input text to create music and audio outputs.
The powers of AudioCraft cover a wide range of audio domains, and it provides MusicGen, which creates music from text using licensed music owned by Meta. On the other hand, AudioGen uses open sound effects to produce audio from text inputs. The experience is improved by the EnCodec decoder’s production of high-quality music with few artifacts.
In addition to its core models, Meta has debuted additional pre-trained AudioGen models that can produce a variety of environmental noises, including footfall, a car honking, and dog barking. With the availability of model weights and scripts that go along with this upgrade, users can be creative and experiment.
Meta’s AudioCraft is an AI-powered tool that might help artists, sound designers, and content creators in finding inspiration and producing songs.
The goal of Meta’s AudioCraft is to make generative models easier to utilise in audio-related projects. Meta aspires to empower creators to explore and push the limits of AI-generated content by making accessible tools and recipes available. This development is consistent with the business’s ongoing initiatives to promote creativity and innovation through technology.