Chapter 22. Meeting 22, Approaches: Agents and Ecological Models 22.1. Announcements • Sonic system reports due and presentations begin: 11 May • Last quiz: Tuesday, 4 May 22.2. Workshop: Sonic System Project Drafts • Last two students present their draft projects 22.3. Agents • Software models of autonomous sub-systems • Complexity and emergent behavior through the interaction of simple agents 22.4. Interactive Music Systems • Computers that musically respond to MIDI messages (control data) • Computers that musically respond to audio (sound through a microphone) • Computers that accompany a human performance based on a shared score • Computer (agents) that musically respond to each other (via audio or MIDI) 22.5. Analysis and Generation • Interactive systems must have two basic components • Components that “listen” to control data or audio information, and decode into musical models • Components that generate musical responses based on analysis 22.6. Interactivity: Theatre • Musical performance is theatre 255• Appeal of technological achievement or drama of technological disaster 22.7. Applications of Multi-Parameter Feature Extraction • Detect articulation, pitch, and tempo and match to a score: score following • Detect articulation, pitch, and rhythms, and build musical responses: interactive systems, installations 22.8. Multi-Parameter Feature Analysis in PD • [fiddle~] object: pitch, event, and amplitude 22.9. Early Historical Examples of Interactive Music Systems • 1967: Gordon Mumma’s Hornpipe (1967): “an interactive live electronic work for solo hornists, cybersonic console, and a performance space”; system analyzes sound form horn and in performance space 256Audio: local (file://localhost/Volumes/xdisc/_sync/_x/eduMitCourses/21m380b/audio/mummaHornpipe. mp3) • 1968: Max Mathews and F. Richard Moore develop GROOVE system at Bell Labs. Real-time performance interface to a predetermined musical score • 1979: George Lewis, with a KIM-1 computer, develops interactive compositions designed to work with improvisation • 1983: Felix Hess creates 40 Electronic Sound Creatures, small mobile machines with microphones and speakers that respond to each other an the envrionment © Felix Hess/John Stoel. All rights reserved. This content is excluded from ourCreative Commons license. For more information, see http://ocw.mit.edu/fairuse.• 1987: Robert Rowe develops a system called Cypher, consisting of a Listener, a Player, and a Critic, used in Flood Gate (1989) 25722.10. Reading: Rowe: Machine Listening and Composing with Cypher • Rowe, R. 1992. “Machine Listening and Composing with Cypher.” Computer Music Journal 16(1): 43-63. • What types of features are extracted during the first level of listener analysis? • What types of features are extracted during the second level of analysis? • How does the chord and key analysis routines work? • What are the three compositional methods employed? • What is the role of the critic? • How is the large-scale behavior of the system varied over time? 22.11. Listening: Rowe • Listening: Robert Rowe, Shells, 1993 22.12. Listening: Ariza • Listening: Christopher Ariza, to leave the best untold, 2009 22.13. Alternative Agent Models • Analogies to human roles • Analogies to ecological models • Analogies to social systems • Analogies to physical systems 22.14. A Model of Particle Feedback Systems • Particles in a dynamic system 258• Particles • Have one or more states, each state with a discrete life span • Particle expired at termination of life span • Life cycle: [('a', 1), ('b', 2)] • Particle Transformers • Have one or more states, each state with a discrete life span • Particle expired at termination of life span • State determines focus of particle • Focus is target state looked for in other particles; transformed with transformation map • Transform map: {'a':[(None, 3), ('a', 1)]} • Related to first order Markov chain • Sensor Producers • Produces one type of Particle • Produces one type of Particle Transformer • Stores a threshold, a target value for a given state • Senses the composition of a collection of Particles • Stores a production count range: given difference from threshold, give a range of Particles to produce (when below threshold) or Particle Transformers to produce (when above threshold). • Production count range: {(-30,-10): [1,2], (1,10): [1, 2], (11, 20): [1, 4], None: [1, 8]} • Environment • Store lists of Sensor Producers, Particles, and Particle Transformers • Provides model of Sensor Producer (one for now) • Provides an absolute discrete value range for sensed particle 259• Specify number of sensors • Can age all Particles by one or more age steps 22.15. Feedback System as ParameterObject • The feedbackModelLibrary ParameterObject :: tpv fml Generator ParameterObject {name,documentation} FeedbackModelLibrary feedbackModelLibrary, feedbackModelName, parameterObject, parameterObject, min, max Description: Produces values from a one-dimensional string rewrite rule, or Lindenmayer-system generative grammar. The terminus, or final result of the number of generations of values specifed by the stepCount parameter, is used to produce a list of defined values. Values are chosen from this list using the selector specified by the selectionString argument. Arguments: (1) name, (2) feedbackModelName, (3) parameterObject {aging step}, (4) parameterObject {threshold}, (5) min, (6) max • A basic model of a Thermostat: particles as heat :: tpmap 100 fml,t,(bg,rc,(1,1.5,2)) feedbackModelLibrary, thermostat, (basketGen, randomChoice, (1,1.5,2)), (constant, 0.9), (constant, 0), (constant, 1) TPmap display complete. • Dynamic age values applied to Particles :: tpmap 100 fml,t,(ls,e,(c,30),0,4) feedbackModelLibrary, thermostat, (lineSegment, (constant, 30), (constant, 0), (constant, 4)), (constant, 0.9), (constant, 0), (constant, 1) TPmap display complete. 260• Climate control: produce both Particles and Particle Transformers :: tpmap 100 fml,cc,(bg,rc,(.5,1,1.5)) feedbackModelLibrary, climateControl, (basketGen, randomChoice, (0.5,1,1.5)), (constant, 0.9), (constant, 0), (constant, 1) TPmap display complete. • Alternative approaches to PO interface? 22.16. Feedback System as