Stability.ai has unveiled the latest version of its AI music generation model, Stable Audio 2.0. Building upon the first version, this new model is capable of creating songs that are up to 3 minutes long (compared to the initial version’s 95 seconds) at a sound quality of 44.1 kHz.
In addition to enhancing text-to-audio capabilities, the model now includes support for audio-to-audio conversion. Users can upload audio files as initial samples and utilize prompts to customize the generated audio in their desired format.
Stable Audio 2.0 has been trained on a dataset of over 800,000 sound files from AudioSparx. Artists with existing works on AudioSparx can opt-out of having their data used for training. Furthermore, Stability.ai has partnered with Audible Magic to address copyright infringement issues arising from AI-generated content.
The Stable Audio 2.0 model is now available for use on the Stable Audio website, with plans to release an API for wider access in the near future. Audiences can also experience AI-generated music 24/7 via YouTube: Stable Radio.
TLDR: Stability.ai introduces Stable Audio 2.0, an AI music generation model capable of creating longer songs at higher sound quality, with added support for audio-to-audio conversion. The model has been trained on a vast dataset and offers users the ability to customize generated audio. Collaboration with Audible Magic aims to prevent copyright infringement. Available for use on the Stable Audio website, with plans for API integration and continuous listening via Stable Radio on YouTube.
Leave a Comment