Automatic design of sound synthesizers as Pure Data patches using Coevolutionary Mixed-typed Cartesian Genetic Programming
1. E N G A G I N G T H E W O R L D
Philippe Pasquier
pasquier@sfu.ca
1
Matthieu Macret
mmacret@sfu.ca
Automatic design of sound synthesizers as Pure
Data patches using Coevolutionary Mixed-typed
Cartesian Genetic Programming
Wednesday, July 16, 2014
2. The Synthesis Calibration Problem
2
Target sound
Sound
Synthesizer
i1
i2
...
in
=
Synthesized
sound
Input
parameters
Wednesday, July 16, 2014
3. Automatic calibration of Sound synthesizers
3
Synthesis
technique Op/miza/on
technique Reference
Addi)ve HMM Heise
(AES
Conven)on
2009)
Substra)ve Par)cle
Swarm
Yoshimura
(Speech
Communica)on
and
Tech.
1999)
Substra)ve Neural
Network Roth
(AES
Conven)on
2011)
FM Cellular
Automata
Serquera
(Applica)ons
of
Evolu)onary
Computa)on
2010)
FM GA
Horner
(Computer
Music
Journal
1993)
Addi)ve GA
Horner
(Computer
Music
Journal
1996)
Granular GA Fujinaga
(ICMC
1994)
Wednesday, July 16, 2014
17. General Sound Synthesis Problem
7
Target sound
Sound
Synthesizer
i1
i2
...
in
=
Synthesized
sound
Input
parameters
Wednesday, July 16, 2014
18. Genetic Programming - Garcia (DAFx 2001)
8
Produc)on
rules
in
had
hoc
synthesis
system
Sound
Synthesizer
i1
i2
...
in
=
Synthesized
sound
Target sound
GP
GA
Wednesday, July 16, 2014
20. Pure Data
• Graphical programming environment for
audio, video, and image processing
• http://puredata.info/
• Popular among musician, artist and
sound designers
10
Wednesday, July 16, 2014
26. 14
Gene Input
1 Input
2
Extra
Param.
Descrip/o
n
Output
Type
3 control control -‐ Division control
4 signal control -‐ Oscillator signal
Wednesday, July 16, 2014
28. Fitness function
• MFCC (Mel-Frequency Cepstral
Coefficients) in windows of 23 ms
16
Euclidean
distance
between
synthesized
sound
and
target
sound
Wednesday, July 16, 2014
29. Fitness function
• MFCC (Mel-Frequency Cepstral
Coefficients) in windows of 23 ms
16
Euclidean
distance
between
synthesized
sound
and
target
sound
Wednesday, July 16, 2014
32. Mixed-typed Cartesian GP
• Integer encoding of chromosome
• Evolutionary Strategy 1+4
• Mutation-only approach
19
Wednesday, July 16, 2014
33. 20
Initialize CGP /GA
population
Evaluate CGP / GA
population
Variate GA
population
Select best CGP
individual
Variate CGP
population
Termination
criteria ?End
No
Yes
Wednesday, July 16, 2014
49. Conclusions
• New approach to automate the design of
sound synthesizers using Coevolutionary
Mixed-typed Cartesian Programming
• Synthesizers represented as reusable Pd
patches
25
Wednesday, July 16, 2014
50. Future works
• Our methods involves many parameters
– Parameter sensitivity analysis
• Extend our system to reproduce not only
one target sound but a set of target sounds
• Improve our fitness function
– Move to a multi-objective fitness function
• Quantitative/Qualitative comparison with
previous similar systems
26
Wednesday, July 16, 2014