Recently, the Stability AI team introduced a new open-source audio generation model named Stable Audio Open. This model's unique feature is its ability to generate stereo audio up to 47 seconds long from text prompts, with a sampling rate as high as 44.1kHz.
Product Entry:https://top.aibase.com/tool/stable-audio-open-demo
Unlike many popular audio generation models currently available, Stable Audio Open's weights are open, meaning anyone can view, modify, and extend this model. This design philosophy not only advances scientific research but also provides developers with more possibilities. More importantly, the model was trained using audio files licensed under Creative Commons, ensuring the legality of the data and avoiding potential copyright issues, reflecting a high regard for ethical data usage.
In terms of technical architecture, Stable Audio Open employs advanced architecture to ensure high fidelity in text-to-audio generation. It can produce high-quality stereo audio, allowing users to enjoy clear and authentic sound experiences. During training, the model was exposed to a diverse range of audio samples, helping it learn richer soundscapes and making the generated audio more authentic and varied.
Additionally, to ensure the new model's performance could rival top industry models, the development team conducted comprehensive performance evaluations. Through the key evaluation metric FDopenl3, researchers found that the model performed well in generating high-quality audio, comparable to other top models in the industry. This comparative study further confirms the superiority and practicality of Stable Audio Open.
The introduction of Stable Audio Open not only focuses on openness and high-quality audio synthesis but also provides a crucial tool for researchers, artists, and developers.
Key Points:
- 🎧 Stability AI has released Stable Audio Open, an open-source model capable of generating variable-length (up to 47 seconds) stereo audio at 44.1kHz.
- 📝 The model was trained exclusively on audio data licensed under Creative Commons, ensuring the legality and ethics of the data.
- 🔍 Compared to top industry models, Stable Audio Open's audio generation quality has been verified to have high fidelity and diversity.