Our AI Music Engine


We provide a realtime AI composer that can adapt to emotional cues, work from existing music and compose music in multiple styles.


Styles

The engine can generate music in over 16 different sub-genres. We group the sub-genres
into larger musical styles, which consist of the following.

Ambient

Our ambient style is designed for functional music to accompany content or activities,
such as open-world games, relaxation, or meditation. It contains multiple sub-genres,
including Electro Ambient, Cinematic Ambient, and Dreamy Ambient.

An example of Electro Ambient in the Happy emotion:




House

Our House music style is designed as a contemporary dance style,
and is the style closest to popular music. It contains multiple sub-genres,
including Electro House, Funky House, Dreamy House, and Minimal House

An example of Funky House in the Happy emotion:




Piano

The AI engine doesn't only allow for electronic styles.
Our Piano style demonstrates this, as it is a simple style with only piano.
The style has sub-genres of Dreamy Piano, Dancey Piano, and Cinematic Piano

An example of Dancey Piano in the Happy emotion:




Rock

Another example of a more analog style is our Rock style.
This style has sub-genres of Classic Rock, Piano Rock, Epic Rock and Electro Rock

An example of Epic Rock in the Happy emotion:

Emotional Dynamics

The engine always has an emotional input, which defines how the music will be composed.
This happens continuously, on the fly

Each of the colored circles represents a particular emotion: Happy, Excited, Tense, Angry, Sad, Tired, Calm, Tender
(from top-right, moving counter-clockwise), and the music changes based on the instantaneous emotional input.

Theme and Variation

Under the hood, the engine generates a musical theme, and then constantly creates variations on that theme.
This means that a new song can be generated at the click of a button

In this video, new themes are generated (called "Seed 1", "Seed 2", etc.)
A theme is always "played" once through before creating variations on the theme

Team



Amélie Anglade is a specialist in Music Information Retrieval–the application of AI to music. Her interests and background are in Music identification, recommendation and discovery. She received her PhD from Queen Mary University in London after doing 2 Master's degrees in Computer Science and Complex Adaptive Systems. She then went on to be the first MIR (Music Information Retrieval) engineer at Soundcloud, built their first recommendation engine, and evaluated their first finger-printing and search engines.




Ryan Groves is a composer, music theorist, and AI Music expert, in addition to being a curator and creator in the XR (Virtual/Augmented/Mixed Realities) space. He did his M.A. in Music Technology at McGill University. He then built the best music app of 2015 with over 6 million users, Ditty. He won the Best Paper award at ISMIR for his work on automated music theory.