Thanks to Violeta Mayoral and Lucia C.Pino for some of the images and videos, to Hangar.org and MTG Music Tecnology Group from UPF Antonio Ramirez and to Befaco.org

MMM

MMM [FLUORESCENT MARKOV BEAT]

MMM [Flourescent Markov Beat] is the first brunch of the MMM series. In a installation/concert format, MMM_FMB with a minimalist and reductionist approach, addresses the rhythmic question and the synesthesia between light and sound.

It consists of an a square array of LED light tubes that turn on and off following a sequence of states generated by a “markov chain model”. This stochastic and “bastard” model is created from the analysis of heterogeneous and diverse folk music rhythms sources. The sound also follows the sequences and it is generated by transduction and amplification of the light and accompanied by digital synthesis.

video excerpt from the performance that happened last april 2021 at the reina sofia national museum in madrid

SOUND GENERATION

In MMM-Fluorescent Markov Beat. The sound is generated in three different ways, these 3 sound layers can be heard alone in some parts of the piece or in combination or juxtaposition with the other layers. The 3 layer sonify the sequence of beats generated by the markov chain model software, and are integrated in the Pure Data software of the performance of the piece.

Layer A:
The sound is generated by transducing the electromagnetic energy of the light through two pick-up microphones (coils) that are in contact with the LED tubes.

Layer B:
This layer plays short samples that come from recording the sound produced with the pick-up microphones on fluorescent gas tubes, which I used in the first prototype of MMM_FMB. I found very interesting, complex and chaotic the sounds generated by this transduction technique with gas fluorescent tubes, but for the new version of MMM Fluorescent Markov Beat I switched to led light tubes, which are more stable and controllable on the larger scale to which the piece evolved.

Layer C:
This third layer is generated by digital audio synthesis and uses a few modulating oscillators to generate low frequency beats.

::INPUTS:: -THE ESSENCES-

For this installation/concert version of MMM-Fluorescent Markov Beat I have used 7 different sound sources of folk music, mainly rhythmic. From different times and locations, there is music from Japan, Malaysia, Peru, the Tuareg culture in Africa, the Inuit of the Arctic, Eastern Europe and Andalusian music from southern Spain,


::ANALYSIS:: -EXTRACTION OF ESSENCES-

In this version a reductionist approach has been chosen, focusing on rhythm and leaving out other musical considerations, for example the timbre. It would be an approach that treats the musical as a symbolic model, in this case the rhythm as a chain of symbols. Rhythm is also treated from a "monophonic" perspective, that is, each beat is analyzed as if it were generated by a single "instrument". In the analysis the interval between each beat (Inter Onset Interval) is calculated in milliseconds with the "Sonic Visualiser" software.


::MODEL:: -SYNTHESIS AND HYBRIDIZATION-

For the creation of the mathematical model that will generate the rhythmic sequence from the heterogeneous data fromthe analysis of the inputs, the software uses Markov chains. The explanation of these chains was developed by the Russian-born mathematician Andrei Markov in 1907. Thus, throughout the 20th century, it has been possible to use this
methodology in numerous practical cases of everyday life. According to Markov, in stochastic (i.e., random) systems or processes that present a present state, it is possible to know their background or historical development. Therefore, it is feasible to establish a description of their future probability.

This model has been programmed in the Python programming language. From the analysis of the intervals of the beats of the inputs we can calculate the matrix of probabilities that a state passes to any of the others in the sequence. To do this the system we have followed is:

*Discrete the intervals (the separation in milliseconds of each beat) in different ranges, assigning each range a number [0,1,2,3,4,5,6,7,8]
* calculate the probability matrix, i.e. the probability of changing from one state to any of the other possible states in the chain, based on the sequences of the inputs, that give us new secuences like this [2,0,1,1,2,3,6,7,8,8,2,0,0,2,3,6,5,…]
* from this probability matrix, sequences can be generated from an initial state value.


I have followed a strategy for "additive" and "evolutionary" compositional structure with the creation of different "models" and their generated sequences.

I call the 7 input sound files [A, B, C, D, E, F, G], from their rhythmic analysis data I have been creating predictive markov models in this order and generating sequences of 3000 steps of each one:

A seq 1 > model(A) > 3000 beats
A B seq 2 > model(A+B) > 3000 beats
A B C seq 3 > model(A+B+C) > 3000 beats
A B C D seq 4 > model(A+B+C+D) > 3000 beats
A B C D E seq 5 > model(A+B+C+D+E) > 3000 beats
A B C D E F seq 6 > model(A+B+C+D+E+F) > 3000 beats
A B C D E F G seq 7 > model(A+B+C+D+E+F+G) > 3000 beats
B C D E F G seq 8 > model(B+C+D+E+F+G) > 3000 beats
C D E F G seq 9 > model(C+D+E+F+G) > 3000 beats
D E F G seq 10 > model(D+E+F+G) > 3000 beats
E F G seq 11 > model(E+F+G) > 3000 beats
F G seq 12 > model(F+G) > 3000 beats
G seq 13 > model(G) > 3000 beats


During the performance these sequences are read (seq1,seq2, seq3, ..) switching on and off the led light tubes and trigering the generation of the sound layers. Pseudo-random variations of speed and tempo are also made to generate a certain organicity and variability in each performance.

One of the ideas behind MMM is to generate hybrid or "bastard" mathematical "models", i.e. models that do not "represent" or simulate a specific style or input material, but that by introducing very heterogeneous and varied sound sources can produce creative musical material and be able to explore new sound experiences and musical perceptions.



SCULPTURE AND DISPLAY:: -CRYSTALLITALLIZATION AND FORMALIZATION-

MMM-Flourescent Markov Beat is materialized in a light structure in the form of a sculpture that adopts different dimensions. This structure consists of four square metal structures of 1.5 x 1.5 meters with 16 LED light tubes each and it is possible to make different configurations depending on the space.

For example, in the performance at the Reina Sofia National Museum in Madrid last April, the four modules were fixed to form a single square of 3 x 3 meters and 64 LED light tubes of 1.5 meters. This square was suspended in the air attached by metal tensor wires to upper bars. And in other performance at Hangar.org Barcelona I used only two suspended modules separated between them, with my setup in the middle.

Software for Markov Model and generation of sequences

electronics assembly diagram

live @Rarefaccìo Hangar.org Barcelona, April 2021

MMM Fluorescent Markov Beat first prototype, developed with the support of the OSIC Grant 2019

Video by Violeta Mayoral at "Sala de espera/Anomia" @Fase, Barcelona 2019