Sep 14, 2023 2 min read

Stability AI launches music-generating AI tool

Stability AI yesterday launched Stable Audio, which it says “will empower music enthusiasts and creative professionals to generate new content with the help of AI”

Stability AI launches music-generating AI tool

AI company Stability AI - best known for its image generation AI tools - yesterday announced the launch of Stable Audio, its first commercial product for music and sound generation.

Stability has been working on AI relating to music for some time, having previously partnered with Peter Gabriel on an animation competition using his music, as well as developing the open source project via Stability Labs.

The commercially available Stable Audio has been trained with music and metadata from production music library AudioSparx and is being very much spun as a tool for music-makers, both hobbyists and professional musicians.

Stability AI CEO Emad Mostaque says: "As the only independent, open and multimodal generative AI company, we are THRILLED to use our expertise to develop a product in support of music creators. Our hope is that Stable Audio will empower music enthusiasts and creative professionals to generate new content with the help of AI, and we look forward to the endless innovations it will inspire".

A free version of Stable Audio can be used to generate tracks of up to 20 seconds based on descriptive text prompts. Meanwhile, an $11.99 a month subscription allows users to generate 500 tracks of up to 90 seconds each month.

Stability AI says that Stable Audio can generate a 90 second track in less than a second and is running on NVIDIA’s A100 GPU, one of the specialised processors used for running AI. This means that a fully utilised GPU could be turning out as many as 86400 generations in a 24 hour period.

With A100s available in the cloud for around $1 an hour, this means the “compute cost” of 500 generations is $0.15 or less. However, it’s more likely that Stability owns its own A100s - which cost around $7000 to buy outright - meaning the true compute cost is probably lower. Of course, the compute cost is only part of the equation, and does not account for the upfront and ongoing costs of developing and training the model behind Stable Audio.

Further hyping up its new product, the company adds: "Stable Audio is ideal for musicians seeking to create samples to use in their own music, but the opportunities for creators are limitless".

That may be true, although since going live the platform has been struggling to deal with demand, resulting in users seeing the message: “Stable Audio is receiving too much traffic right now - please try again in a few minutes”. So, you know, there are some limits.
Great! You’ve successfully signed up.
Welcome back! You've successfully signed in.
You've successfully subscribed to CMU.
Your link has expired.
Success! Check your email for magic link to sign-in.
Success! Your billing info has been updated.
Your billing was not updated.
Privacy Policy