In major news for global audio production, Stability AI, the company behind stable diffusion, an artificial intelligence-based art generator app, recently launched an open model for creating sounds and music. What Google calls Stable Audio Open means that their generative model is capable of producing drum beats, instrument riffs, ambient noises, and other ‘production elements’ of videos, films, and TV shows out of a text description.
Features
The innovative model, which can create audio tracks up to 47 seconds long, was trained using about 486,000 samples played from FreeSound and the Free Music Archive. This means that the sounds that are produced from the Stable Audio Open are fully and completely free from any royalty, making them a very useful resource for many creators and producers in the audio platform.
Stability AI has also highlighted the potential for users to fine-tune the model on their custom audio data. For example, a drummer could fine-tune the model using samples of their own drum recordings, allowing them to generate new beats with a unique, personalized sound.
Limitations
However, Stable Audio Open is not without its limitations. The model is not currently optimized for producing full songs, melodies, or vocals, and its terms of service prohibit commercial use. Additionally, the model may not perform equally well across musical styles and cultures or with descriptions in languages other than English, due to biases in the training data.
Stability AI, an organization that was previously involved in controversies relating to its position on the utilization of copyrighted content in AI training, emphasized that Stable Audio Open was trained with only disentangled, royalty-free content. One may have found these concerns to practice preventive action by releasing the model and pushing its paid-for services, such as the Stable Audio service.
Why Stability AI CEO Emad Mostaque Resigned
With the increased number of music-generating AIs on the market currently, the question of ownership over generated music or the possibility of misuse of the system is coming to light. In May, music giant Sony Music tried to sound the alarm by taking the uncharacteristic step of mailing 700 AI firms, warning them against the ‘unauthorized use’ of its content for training its audio generators. However, in March, the first state law on AI in the United States was enacted in Tennessee to help counteract abuses in the music industry.
With these issues in mind, Stable Audio Open fits in as a new release for the world related to audio production, which is a push in the right direction as it offers creatives and professionals an effective, flexible, and royalty-free tool.
In conclusion, Stable Audio Open is a highly effective AI sound mixer developed by Stability AI that opens up a wide variety of opportunities for numerous target audiences. Despite some drawbacks and present questions regarding copyright in AI-produced music, the availability of royalty-free audio samples and the ability to adjust the model according to one’s preferences can be regarded as the major pluses for the audio community. With further advancements in the field of artificial intelligence, it will be more and more interesting to observe how the entire field of music and sound production will be impacted.
Stability AI Introducing Stable Video Diffusion: A Leap in Generative Video AI
