Stability AI has introduced Stable Audio 2.0, the latest version of its text-to-music generation system, designed to assist artists and musicians with an expanded array of creative tools. According to the company’s announcement on Wednesday (April 3), this upgraded version empowers users to craft full-length music tracks featuring traditional song structures and superior audio quality through natural language prompts.
Following the acclaim garnered by Stable Audio 1.0, released in September last year and recognized as one of TIME’s Best Inventions of 2023 for its ability to generate short audio clips from textual descriptions, the new iteration builds upon its predecessor. It enables users to create complete songs of up to three minutes in length at 44.1 kHz stereo, facilitating a broader spectrum of musical compositions, from instrumental pieces to structured arrangements encompassing introductions, developmental sections, and conclusions.
In a blog post, Stability AI remarked, “Stable Audio 2.0 establishes a new benchmark in AI-generated audio,” highlighting the introduction of audio-to-audio generation, which allows users to upload their own audio samples to influence the style and sound of AI-generated outputs. Alongside the extension in duration, Stable Audio 2.0 introduces various features including style transfer for aligning generated or uploaded audio with existing tracks, sound effects creation, and generating variations.
Addressing concerns surrounding copyright issues in AI development, Stability AI emphasized that Stable Audio 2.0 was exclusively trained on a licensed dataset from the AudioSparx music library, respecting opt-out requests and ensuring fair compensation for creators. Additionally, the update integrates Audible Magic to scan audio uploads for potential copyright infringement, providing real-time content matching to prevent violations.
Stability AI positioned Stable Audio 2.0 as one of the most potent and adaptable generative AI music tools, enabling musicians, producers, and creators to utilize AI collaboratively for music composition, audio experimentation, and content creation in unprecedented ways. Technical insights into the model’s architecture were also provided, elucidating its capacity to recognize and replicate the intricate structures essential for producing high-quality musical compositions.
The new model is accessible for free on the Stable Audio website and will soon be accessible via the Stable Audio API. Furthermore, Stability AI has launched Stable Radio, a continuous live stream showcasing tracks generated by Stable Audio.
