Stability AI's audio generation model creating high-quality music and sound effects from text prompts up to 3 minutes.