AI that generates music according to conditions
Our music generation system can automatically compose and generate music by specifying the genre and characteristics of a song. The a number of control parameters can be set and changes will be reflected in real time, allowing nuanced operations to be dynamically performed in response to inputs such as body movements, biological data from sensors or the environmental conditions of a particular location.
Feature
Instantly generate MIDI signals
Instantly generate MIDI signals by specifying given music genres, song characteristics, etc.
Real-time control
Real-time control of tempo, instruments, number of notes, speed of development, etc. on a measure-by-measure basis
Feature
Feature
Generate music that changes according to biological data, in environment
Possible to continuously generate music that changes according to biological data from sensors, changes in environment etc.
Use Case
Functional music
With the goal of enhancing human functions such as concentration and performance during exercise, we can provide dynamic music applications that change the content generated in real time in response to biological responses.
Music for Safe Driving
According to the driving conditions of the car, we can provide music that continue to change. In the future, it can also be utilized as in-car entertainment when autonomous cars become widespread (refer to the top video for more information).
Music that shapes the environment
Music can be generated in hotels, offices, and other living environments to suit the situation at any given time. It is copyright-free and can be used without restrictions on media or location.
Client
This product was used for Suntory’s special tea campaign website for a project to generate music that matches the user’s diet.
AI music generation site based on dietary data: SUNTORY TOKUCHA MUSIC (https://tokuchamusic.jp/)
This product was used in a BGM generation system for Shiseido’s first flagship store, “SHISEIDO GLOBAL FLAGSHIP STORE”.
”SHISEIDO GLOBAL FLAGSHIP STORE” opened in Ginza, Tokyo: Interior view of the store
Technology
A deep learning model based on Transformers and Recurrent neural networks for generating MIDI signals is constructed by learning from various musical pieces. The architecture is configured to accept not only changes in initial conditions but also changes in conditions during playback. The actual tone of the music played is selected from a synthesizer, sampler, or other sound source. Research has shown the positive effects of music on health, but how music specifically affects particular biological responses is an area that requires research and development on a case-by-case basis.
Tech Spec
Price System
Licensing period: Monthly Developer’s license: Yes
Input/Output
Input: Composition parameters (genre, tempo, instrument, number of notes, development, etc.) Output: MIDI signal or WAV
Operating Environment
Cloud computing: Standard API provided On-premise environment: Possible by consultation
Processing Speed
Real-time
Other products
