Meta Platforms, on Wednesday, unveiled its latest innovation, AudioCraft, an open-source AI Tool designed to enable users to create Music and audio using text prompts. This AI tool comes bundled with three distinct models, namely AudioGen, EnCodec, and MusicGen, each catering to various aspects of music, sound, compression, and generation, according to Meta.

The MusicGen model is trained using Meta's proprietary and specifically licensed music, ensuring a unique and diverse musical output. However, concerns have been raised by artists and industry experts regarding potential copyright violations, as machine learning algorithms rely on recognizing and reproducing patterns from web-scraped data.

As per Meta's blog post, the MusicGen model generates music based on text prompts, while AudioGen focuses on creating audio from text prompts. Additionally, Meta has introduced an improved version of their EnCodec decoder, which contributes to generating higher-quality music with fewer artifacts. The pre-trained AudioGen models allow users to generate environmental sounds and sound effects, such as dog barks or vehicle sirens.

Researchers and practitioners will have access to these models to train their own datasets, and Meta assures that these models can produce high-quality audio consistently over extended periods. Developed internally over the years, AudioCraft's models are intended to become valuable tools for musicians and sound designers, with continuous improvements and improvisations based on user feedback in the pipeline.

Earlier this year, Alphabet, Google's parent company, introduced its experimental audio-generating AI tool called MusicLM. As the AI technology continues to evolve, such innovative tools are paving the way for novel and exciting possibilities in the world of music and audio creation.