AudioMatt

antiloop-m

AntiLoop M

Barak Chamo

Glass Punk is an algorithmically-composed musical performance inspired by the works and style of Phillip Glass and Daft Punk. 

A Markov-chain (sequential probability model) was trained on a modified corpus of MIDI scores by both Glass and the French duo that's been adapted to a consistent key and mode, resulting in a state-machine that fuses both musical styles into one. Different pieces were used to train multiple chains for use as MIDI sources for bass lines, melodic and harmonic progressions.

The piece was performed live with Max/MSP and Ableton Live, using Live to trigger clips and control the playing of parts and rhythms and feeding MIDI "pings" into the Markov Chain to receive note values. The result is a composition that is rhythmically coherent and develops over time, yet it's melodic and harmonic content is indeterministic. This process yielded interesting results that are reminiscent of source materials by all musicians, a dynamic and ever-developing hybrid.

Glass Punk is an algorithmically-composed musical performance inspired by the works and style of Phillip Glass and Daft Punk. 
A Markov-chai

glass-punk

Glass Punk

aengus

This is an object for Max that implements variable order Markov models for generating melodies, beats or other musical data. It can be used for Markov models of any order. The download includes a version of Pachet's Continuator for Max. This is a variable order Markov model-based algorithm that can learn a musician's style (to some extent) and trade licks in real-time- lots of fun!

The source code for the VMM object is available on github. Download the VMM object here.

Please leave questions in the comments section below.

variable-order-markov-model

Variable Order Markov Model

Anthony Palomba

Tired of trying to figure out markov states changes in your head? Then this handy utility is just what you need. It reads and writes data sets (text files) that contain state information as well as data that corresponds to those states. Once a data set is loaded, you can then trigger a new state with prob and the corresponding data element will be output. You can also manually enter in the states and data.


markov-utility

Markov Utility

Matthieu Pernaud

as i learnt markov chains with pure data with [float] objects and the [moses] object, i designed a simple abstraction that acts exactly the same as [moses] in pure data.

Search Results