Stable Audio

Stable Audio is an AI-driven platform that enables users to generate original music and sound effects using text prompts and audio-to-audio transformations. The platform caters to musicians, sound designers, and content creators, offering tools for both beginners and professionals.

Key features:
- Audio-to-audio generation: Transform uploaded audio samples into various sounds using natural language prompts
- Text prompts: Guide the AI model towards desired output by describing how the audio should sound
- Adjustable input audio and prompt strength: Control the level of modification and balance between the input audio and text prompt
- Multi-track generation: Create full tracks with coherent structures, including intros, developments, and outros
- Stereo sound effects: Generate high-quality, structured sound effect compositions

How it works:
1. Upload audio samples or use existing AI-generated audio
2. Add a text prompt describing the desired audio output
3. Adjust input audio strength and prompt strength parameters
4. Generate the audio based on the input and parameters
5. Refine the output by iterating and adjusting parameters

Integrations:
- AudioSparx: Exclusive training data for Stable Audio's models
- Freesound and Free Music Archive: Training data for Stable Audio Open

Use of AI:
Stable Audio leverages artificial intelligence to generate original music and sound effects based on user input and adjustable parameters. The platform offers an intuitive interface for users to interact with the AI models, upload audio, and fine-tune their output.

AI foundation model:
Stable Audio 2.0 is built on the Stable Audio AudioSparx 2.0 model, which utilizes a latent diffusion architecture. This model employs a highly compressed autoencoder and a diffusion transformer to generate high-quality, full-length tracks with coherent structures.

How to access:
Stable Audio is available as a web app, accessible through a browser. The platform offers different pricing tiers, including Studio, Creator, and Enterprise, each with varying levels of track generations, track duration, and monthly upload amounts. Additionally, Stable Audio Open is an open-source model optimized for generating short audio samples, sound effects, and production elements using text prompts.