previous notes:
patches/osc_plus_pipes.pd patches/countersusw.pd patches/space.pd patches/i_am_a_neuron.pd patches/in-out.pd patches/freq1.pd patches/pink~.pd patches/archiver.pd patches/multiplex.pd patches/mix.pd patches/thresholds.pd
to http://1010.co.uk/plenum_patches3.tar.gz
these are the components :- modules like jekyll will make use of triggers and counters and archiver. need overarching connectivity app which shows all send and receive
.pdrc::
-lib zexy -path /usr/lib/pd/externs/ -path /usr/lib/pd/iemabs -path /usr/local/lib/pd/extra/ -path /root/xxxxx/OSCx/OSC -lib OSC -path /usr/local/lib/pd/externs -lib iemlib1:iemlib2:iem_mp3:iem_t3_lib -lib /root/xxxxx/plenum/bin/hammer -lib /root/xxxxx/plenum/bin/sickle -lib /root/xxxxx/plenum/ggee-0.25/gcanvas -lib /usr/lib/pd/extra/pipewrite~ -lib /root/lunch/ext13/piperead~ -lib /root/externals/ann/src/ann_som
+ add chaos: -lib /root/pd/pdstuff/pd-externals-20030311/chaos/chaos
For Plenum setup:
1) we want 8 mics 1-8 input pass-through
set now as default hdspmixer file
2) initialisation script:
#!/bin/bash /usr/bin/hdsploader & sleep 4 /usr/bin/hdspmixer & sleep 1 /usr/bin/jackd -d alsa -d hw:2 & sleep 4 /usr/bin/qjackctl &now aliased in .bash_profile to:
rmesetup (tho. hdpsloader is sometimes tricky)
and pd:
/usr/bin/pd -jack -inchannels 12 -outchannels 12 -open /usr/lib/pd/doc/3.audio.examples/A07.frequency.mod2.pd -r 48000 &
to:
pdrme
3) for patching lineout is to [dac~ 9 10]
see software
http://1010.co.uk/plenum_tools02.tar.gz
~/xxxxx/plenum_tools/oscdump 5252 > ~/pdfifo
and butchered test2-osc-sc patch as tester (though some fine tuning needed)
such test patches can be enlarged to module patches
is under ggee/help-mixer.pd
basic jack/pd/hdsp setup for plenum
1) 8 mics pass through all 2) 8 inputs -> pd 3) 2 outputs to 9 and 10 (line out/aux/whatever called in pd) 4) samplerate would be nice bring down to 8000 in pd i guessinput and output all abstractions with [s] and [r] or [s~] [r~]
some sort of naming scheme - [r data_act1_blah]
seperate modular patch for each act. string together for acts/changes
so 5 acts/modules as we have below.... + in, out, piping, OSC interface and mixer (10 patches) - maybe 11 with an archiver patch
unsig~ : signal to float (round off (with round~) and probably work more tabread~ and write~) pink~ : pink noise t3_bpe time tagged trigger break point envelope t3_delay time tagged trigger delay t3_metro time tagged trigger metronom t3_timer time tagged trigger timer ------------------ t3~ - time-tagged-trigger -------------------- -- inputmessages allow a sample-accurate access to signalshape -- t3_sig~ time tagged trigger sig~ t3_line~ time tagged trigger line~
and generic pd reference:
trigger [t] eg: [t f f] outputs input from right to left converting to say f for float, s symbol etc. time: ------------------------------ TIME ---------------------------------- delay - send a message after a time delay metro - send a message periodically line - send a series of linearly stepped numbers timer - measure time intervals cputime - measure CPU time realtime -measure real time pipe - dynamically growable delay line for numbers threshold~ detect signal thresholds snapshot~ - sample a signal (convert it back to a number) bang~ - send a bang message after each DSP block readsf~ - soundfile playback from disk writesf~ - record sound to disk tabsend~ - write one block continuously to a table tabreceive~ read one block continuously from a table delwrite~ - write to a delay line delread~ - read from a delay line vd~ - read from a delay line at a variable delay time
and zexy::
lp write to the (parallel) port (linux only) matrix operations
most culled from:
http://lists.puredata.info/pipermail/pd-list/2004-11/023740.html
b) plenum - pd only spatialisation and neuron model [settle on ann_som for latter, spatial ->
pd -lib bin/hammer -lib bin/sickle -lib ~/ggee-0.25/gcanvas -open ~/plenum/spatialisation-help.pd
as before but we can obviously modify for multiple mics - also some easy set up of multiple canvas/overlap - way of marking different mics
neuron attraction/repulsion. boxing club is chemical zone
neuron as own pd extension
signal strength -> exhaustion/trigger time -> overarching weight -> //properties of -- neuron
across vector. neuron fires - oscillation network. connected in a complex net and also to inputs
modules as specified by plenum desc::
1) i_am:
Exploration of neuronal, neural network models as triggering mechanisms for dictated software events. One neuron or many maps the room, a map of neurons, a network for raw data, for piping, as raw as possible. Neural net thing is mapped itself over participants. There must be some way for participants to be mapped over the net and re-enter the net simulated. Self entry needs to be coded, raising the issue of interface. Again, it's unclear which side of the interface we're discussing, as the social is mapped to the neural as a question of elements and the individual: group as neuron group, trigger over limit can be made unresponsive - thus reverse louder participant, drained neuron, overexerted neuron, waiting and empty neuron partnering.
summary: neuronal, triggering components: pd: ann_som, bonk~, fiddle~, threshold~. cl: i_am
2) alice:
The script is interrogated and opened up to ridicule. The code is full of (rabbit) holes and sample transformations. Our pleasant neural system now accepts wild triggerings. A tea party.
summary: PLT Scheme -> speech components: flite and flite pd external. OSC tools and connections
3) jekyll
Jekyll software modules should dominate the room. Commandline modules are patched together within a stacked and tense atmosphere which will arrive to duel. Jekyll assumes sampling or data slicing a la slow scan image-sound, piping and leaked and labelled containers. Timeslice audio laid open to inspection. As from a distance, from within the train carriage, distant objects maintain their place, whilst those closest rush past, a similar spectrum will be attempted. In addition the exterior scene can be moved with no train motion. Time slicing and distance. Multiple versions of the same software will be showcased with reference to the multiple edits and censorsnips concerning Docteur Jekyll et les Femmes (1981). Walerian Borowczyk.
summary: slicing, over-recording and selection components: commandline: piping, jekyll, OSC tools. pd: sampling and triggering to explore further. soundfiling capablities?
4) H bomb
Simulation is made evident through noise and spatialization. The audience is in the patch/code. Code audio is broadcast into the audience simulating space; where are the microphones recorded in the machines placed in simulated space and now heard in the room through two speakers?
Simulation at entry point/interface into the machine. The participants in our expanded software must be mapped into the computer model. Where they meet is the (ugly) interface. What the machine can know is important for its mapping. For example if the microphones are moving the machine doesn't know where they are (unless we code in very complex algorithms that would compare background noises - simulating space is easier than reading space) - unless we at the interface say so - ie. we humanly trace it - or we use some machine vision or some kind of interference pattern. Noise or waveforms suggest themselves.
summary: spatialisation, frequency generation, noise - interference patterns (chaos theory) components: pd hammer and sickle stuff. noise? ap. chaos - again see roessler etc. in pd
5) pink light.
one waveform.
summary: one waveform components: how is it generated?
each modules expands on the previous - but we need architecture of abstraction - pd abstractions
thus:
8 mic neuron patch using bonk~ or fiddle~ or threshold~ -> pd
alice speech synth - scheme ->pd/osc/flite
timeslicing and cross-talk piping all+commandline
spatialisation and noise-interference - i_am + some spatial pd patch in concert with neurons
waveform - sc3?
map of neurons. network.
attempt in few days command line nn piper and OSC piper.
as jekyll -( can be similar to one pixel sampling - data slicing - a standard for data slicing), alice - scheme layer
5 softwares, 5 acts - last as pink light patch
further/opening module could well be substance
-- interface
elements to resolve:
1) triggered speech recording, archiving and seeking
2) notion of timeslicing
3) neural net mapping - audience mapping
4) leads to spatial/audience simulation - spatialization
5) interface
6) exchange of messages/data with other Pd'ers
7) 8 mics cross mixer network
pd or smaller scheme/commandline apps? or both. pd scheme interface: http://www.westnet.com/~lt/pd/pd-scheme.html
also with pd for xxxxx_on_a_wire:
1) ap->pd pd->OSC experimental patch
2) sniffer-OSC stuff
(again maybe see OSC in scheme also - http://www.artm.org/pub/lisp/osc/ )
towards more totally integrated language. emacs (emacs with PLT, SIOD - pd and REPL tied?). alice framework !
pd for audio abstraction + interface + some dsp shell for neural net. spatial. osc elaboration - shell object in pd scheme and emacs as language interface, integration
plan one day scheme/pd/emacs storm
also installed cyclone external for hammer and sickle
1st plenum notes:
The PLENUM patch is concerned with interface which operates in two directions - from the realm of speculative or expanded software (aka the social) into the reduced realm of codified software and (as output, as interference, as noise) the entry of executed results back.
The patch is thus concerned with that which can be exchanged (or mapped over) on this interface, that which is knowable for the patch - time, sound as data (amplitude and thus frequency which then opens up to statistical analysis across various domains), and (as we will have 8 mics distributed in space which input into RME multiface) space.
The Pd patch will operate some kind of triggered switching/time slicing between participants and echoing in escalatory fashion. It also needs to be readily extendible (heavily abstracted and based on message passing). It is concerned with an escalating overmapping of expanded and reduced software domains. The problem states itself as that of the practical. Substance.
2nd:
I'll check out Yves' external - i think it could be of interest in that it is concerned with what he calls "audience simulation" which maps onto our/my possible concern with mapping - mapping in/as simulation at entry point/interface into the machine. the participants in our expanded software must be mapped into the pd patch/computer model. where they meet is the (ugly) interface. what the machine can know is important for its mapping. again a bit abstract. if the mics are moving the machine doesn't know where they are (unless we code in very complex algorithms that would compare background noises - simulating space is easier than reading space) - unless we at the interface say so - ie. we humanly trace it - or we use some machine vision
or some kind of interference pattern
[or we can give more data through sensors]
so maybe space becomes simulated a la yves - we/participants maintain and exchange mappings and simulations ahem!
and space knowledge becomes knowledge of difference - we know we have 8 differing mikes which are located in differing places (they cannot be in the same place). pd patch philosophy
yes. i propose eight (or x but we should fix) pd-ers with 5 patches each - one for each act and maybe varying rulesets overarching each patch/set of pd-ers
overmapping of spatial or rather simulationas we don't know where is the mic
pd -lib bin/hammer -lib bin/sickle -lib ~/ggee-0.25/gcanvas -open ~/plenum/spatialisation-help.pd
(using yves audience~ is very crackly)
from gaim chat with j:
neural net thing is closer mapped itself over participants "i am a neuron"
(20:28:04) mar: next question is how - how participants can be mapped over the net and re-enter the net simulated
(20:29:26) mar: or entry sound switching - each set of samples assigned correctly or incorrectly for flow of participants
modelled neurons (limit to trigger) can be assembled in changing networks
varying algorithms to simulate networked triggers
group as neuron group
trigger over limit can be made unresponsive - thus reverse louder participant.
drained neuron
overexerted neuron
waiting and empty neuron partnering
at the same time timeslicing or some sort of holographic neuronal model
started on PLENUM pd patch:
using spigot~ extension from yves for signal routing:
http://ydegoyon.free.fr/software.html
some PLENUM notes:
timeslice audio a la jekyll
store and recall slices triggered perhaps by threshold~
also maybe making use of line~
switch inputs - 8 pre-amped mics into RME
audioline scan::
slow scan scanning participants audio timed expectation sample segments assigned to participants p. neural net mapped over p. switching noise ratio contributive noise ratio ratio. per slice to switch - threshold~ and choice to switch/escalate/flood - delay feedback to switch/deny other p. RME multiplication and conditions of cross-talk time slicing and distance relative movement