Ernst

Ernst is an automatic improviser that re-uses the sounds produced by the user to make music. The input from the microphone is recorded and segmented into several snippets of audio. These snippets are analyzed for their pitch, volume envelope, rhythm characteristics and metrical stage within a larger musical phrase. This information, along with its relation to the surrounding snippets, is fed into 4th order Markov chains that are responsible for outputting responses to the audio input. As soon as the characteristics aimed for each response sound are calculated, the system searches for the recorded snippet that more closely resembles these features and triggers it after all the necessary processing (e.g. pitch and time-shifting, envelope morphing).

Download

Macintosh

Windows