Music generation using neural networks is a promising approach for game developers, as it allows for creating dynamic and interactive music possible to be adapted to the gameplay. The paper presents a music generator designed for computer games that utilizes artificial neural networks and MIDI compositions for games as a training set. The generator is targeted at low-budget projects and independent game developers. It is a complete Internet application that enables music generation of specified positive or negative sentiment. It applies procedural composition elements and enables manual tempo and pitch range setting. The generated music is presented as a piano roll. In editing mode, the user can change the pitch of an individual note, add and remove some notes, and add some sound effects. It is also possible to input a starting piece of melody that the generator will further develop into the full track.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=22077
(348KB)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
