Meta's New AI Tool Lets You Make Music and Audio from Text


Meta's New Open-Source AI Tool Lets You Generate Music and Audio from Text

Artificial Intelligence (AI) has indeed established its presence, as tech giant Meta introduces its latest tool, AudioCraft. This tool can generate "high-quality and realistic" music and audio by simply using text prompts, which further solidifies the permanence of AI in the field of technology.

According to a press release, AudioCraft will aid professional musicians in exploring unique compositions more easily or help business owners create background music for their advertisements, among others.

The new AI-powered tool will open-source three models: MusicGen, which composes music from text prompts; AudioGen, which is trained to create sound effects; and an upgraded version of Meta’s EnCodec, which makes compressed but high-quality music. With these, AudioCraft is pegged to be a one-stop shop for generating, composing, and compressing music and audio.

Meta likewise mentioned that AudioCraft is a significant movement in the AI space, as generative AI tools today foray more into creating videos, images, and texts. “There’s some work out there, but it’s highly complicated and not very open, so people aren’t able to readily play with it,” Meta said.

Meta also likened the development to the invention of synthesizers, referring to AudioCraft’s MusicGen model as “a new type of instrument.”

Currently, a demo lab is open to the public via its website, where users can access all three models. To learn more about the new tool, you can also visit Meta’s AI blog here.

Get the latest curated content with The Beat Asia's newsletters. Sign up now for a weekly dose of the best stories, events, and deals delivered straight to your inbox. Don't miss out! Click here to subscribe.