Stability AI Unveils Advanced Music Generation Tool with Extended Capabilities

Stability AI, a prominent artificial intelligence (AI) research company headquartered in London, has recently announced a significant update to its music generation platform, Stable Audio. This upgrade, designated Stable Audio 2.0, represents a notable advancement in the field of AI-powered music creation.

Stability AI positions Stable Audio 2.0 as establishing “a new standard in AI-generated audio.” The core improvement lies in its capacity to generate high-fidelity, complete musical pieces with a coherent structure. Notably, the maximum track length has been doubled from the original version, enabling the creation of compositions up to three minutes in duration, all delivered in high-quality 44.1kHz stereo.

Beyond its text-to-audio functionality present since launch, Stable Audio 2.0 introduces a novel audio-to-audio transformation feature. This feature empowers users to upload copyright-free audio samples and utilize natural language prompts to creatively transform them into a diverse array of sounds. Stability AI has provided a demonstrative video hosted on YouTube showcasing the potential of this function in fostering the generation of novel music.

In a blog post published on their website, Stability AI emphasizes a key differentiator of Stable Audio 2.0: its ability to produce extended compositions with structured elements, including introductions, developments, outros, and even stereo sound effects. This enhanced capability positions Stable Audio 2.0 distinctly within the landscape of cutting-edge AI models.

Furthermore, Stability AI is offering unrestricted access to Stable Audio 2.0 through their website, free of charge. This move allows anyone to experiment with the platform and explore its creative potential.

The launch of Stable Audio 2.0 coincides with ongoing concerns within the creative industries regarding the potential impact of generative AI on human talent. This is further amplified by a recent open letter co-signed by prominent artists including Billie Eilish and Nicki Minaj, advocating for safeguards against the “predatory use of AI” replicating the voices and likenesses of creators without authorization.

Stability AI acknowledges these concerns and underscores in their blog post that Stable Audio 2.0 was exclusively trained on a licensed dataset from the AudioSparx music library. This dataset encompasses over 800,000 audio files and adheres to opt-out requests, ensuring fair compensation for the contributing artists.

Back to top button