Sony Group has created a technology capable of detecting the underlying musical works used in AI generated songs, potentially enabling copyright holders to claim compensation when their compositions are utilized in machine learning and content creation. (Reports Nikkei Asia)
The system analyzes which musicians’ songs were employed in training AI models and generating new music. It quantifies the contribution of each original work, producing outputs indicating percentage based influence from specific artists or catalogues.
When AI developers agree to cooperate, Sony obtains data by connecting to the developer’s base model system.
In cases where cooperation is not attainable, the technology estimates original works by comparing AI generated music against existing compositions.
The development arrives as the AI boom has sparked numerous disputes over unauthorized use of copyrighted music, video, and text for machine training.
AI generated songs mimicking the voices of well known singers have circulated online, raising questions about rights holder compensation.
Sony envisions the technology facilitating a system that distributes revenue generated by AI music to original songwriters based on their calculated contribution.
The technology was developed by Sony AI, the company’s research and development division. A related paper was accepted at an international conference. The division has also developed methods to prevent AI from imitating animation styles and characters, such as those from Studio Ghibli.
These technologies can be applied to videos, games, and characters. Sony Group stated it has not yet determined when the system will be deployed commercially.
Sony envisions AI development companies incorporating the technology into their own models, and content companies using it during license negotiations.
