grh.mur.at - Pd

How to become a Pure Data / GEM Professional

grh — Sun, 20 Jun 2010 18:31:49 +0000

These are materials from a lecture about realtime audio and video programming (using GEM and Pure Data) at the University of Applied Science, FH Joanneum, for the master programme Media and Interaction Design.

Advanced Psychoacoustics 2: Silent Subliminals

grh — Fri, 18 Sep 2009 14:05:04 +0000

Do you want to influence a humans behavior ?
Or maybe induce some thoughts in the brain of all people listening to a specific radio channel ?
Then you found the right blog entry, which will present you a Pure Data program to change the attitude of your desired audience ...

Sound Spatialization in Pd

grh — Wed, 24 Jun 2009 18:22:51 +0000

Year:

2007

Authors:

Georg Holzmann

Type:

Workshop

Publisher:

Proceedings of the Linux Audio Conference 2007

Abstract:

The goal of this workshop is to show how to position sound in space (stereo, multichannel and binaural). This should be done from a user point of view, without explaining the detailed mathematics behind the algorithms. Therefore existing and open-source implementations in Pure Data will be used and explained.

Topics:

stereo-panning methods
vector based amplitude panning (VBAP)
ambisonic
binaural ambisonic and 3D room simulation

To all topics I will explain the handling of the Pd implementations and the advantages/disadvantages of the specific methods, demonstrated on examples.

All workshop materials can be downloaded below.

Media:

Pd spatialization workshop

Genetische Algorithmen in Komposition und Computermusik

grh — Wed, 24 Jun 2009 18:10:35 +0000

Year:

2003

Authors:

Georg Holzmann

Type:

Technical report

Publisher:

Internet Publication

Abstract:

Meist werden komplexe Systeme von Algorithmen in der algorithmischen Komposition
verwendet. Dadurch entsteht eine Vielfalt an Parameter, die die intuitive Steuerung solcher
Systeme erschweren.
Mit Hilfe von Interaktiven Genetischen Algorithmen (IGA) kann man Variationen dieser
unzähligen Parameter nach eigenen ästhetischen Vorstellungen „züchten“, ohne ein Wissen
über die darunter liegende Struktur haben zu müssen und hat trotzdem noch ein hohes Maß an
Kontrolle.
Diese Arbeit bietet einen Überblick von Anwendungen Genetischer Algorithmen in der Musik und präsentiert eine neue Implementation.

Publication:

Genetische Algorithmen in Komposition und Computermusik

VideoIO

grh — Wed, 24 Jun 2009 15:23:31 +0000

Short Description:

general video IO framework for Pure Data

Started in:

2007

Authors:

Thomas Holzmann

Authors:

Georg Holzmann

License:

GNU General Public License (GPL)

Programming language:

C, C++

Overview:

This framework should provide a general video input and output system (reading/writing various files and from/to devices) for Pure Data. The interface is platform independent and the needed libraries are checked and loaded at runtime, so the host application (e.g. Gem, Gridflow, PDP) has no compile time dependency on all those video libraries.

At the moment only the interface to Gem is implemented and consists of the following objects: pix_file_read, pix_file_write and pix_devic_read. They use the gstreamer plugins for all these opterations, but alternative libraries, especially on Windows and OSX, should be added in future.

What they can do ATM:

pix_file_read: reading audio and video from all formats gstreamer supports (quite a lot), reading from dvd, http, mms, udp streams, ...
pix_file_write: writes in some codecs to disk, also streams over udp
pix_device_read: can read from camera and DV devices

This project was part of the Google Summer of Code program 2007.
The latest source can be downloaded from the SVN repository, an other (maybe outdated) version can be found in the download section below.

Version Control System URL:

VideoIO SVN repository

Release Tarball:

VideoIO source and documentation

sndfiler

grh — Wed, 24 Jun 2009 14:48:13 +0000

Short Description:

a threaded soundfiler for PD using libsndfile and libvorbisfile

Started in:

2005

Authors:

Tim Blechmann

Authors:

Georg Holzmann

License:

GNU General Public License (GPL)

Programming language:

Overview:

A threaded soundfiler for Pure Data using libsndfile for reading uncompressed and libvorbisfile for reading OGG files.

FEATURES:

threaded reading of multichannel soundfiles into arrays
threaded resize of arrays
support for OGG

REQUIREMENTS:

pd >= 0.39
libsndfile
libvorbisfile
threadlib

For more information look at the help patch.
See the file INSTALL for installation instructions.

Version Control System URL:

sndfiler SVN repository

threadlib

grh — Wed, 24 Jun 2009 14:36:37 +0000

Short Description:

C and PD library for threaded patching and threaded PD externals

Started in:

2005

Authors:

Georg Holzmann

License:

GNU General Public License (GPL)

Programming language:

C, Pd

Overview:

C and Pure Data library for threaded patching and threaded PD externals, heavily based on pd_devel code by Tim Blechmann (his detach/join for pd_devel).

PD objects for threaded patching:

detach: run part of the patch in a helper thread
join: synchronize messages to pd's main thread
sleep: blocks the system for a specific time

WARNING:
These objects (especially detach/join) are very experimental and may crash your patches if you use them in a wrong way, because some externals/internals of PD are not threadsafe !!!

Features for PD external programmers:

lockfree FIFO from pd_devel
callback system like in pd_devel: instead of the idle callbacks, which are not really usable in current main pd, it uses clock callbacks
USAGE: you have to link your externals against threadlib - see sndfiler as an example
ADVANTAGES: so it's possible to use the lockfree FIFO and sys_callback in main pd (and the same source code for main and devel pd)

Many thanks to Tim Blechmann for his code and help!

detach:
Detach is working on control objects only, creates a helper thread and runs all functions of the following objects in this helper thread. If a message from that thread reaches a join object a callback for running the outlet function will be places in the scheduler and run in the next scheduler loop - so you can synchronize the message with pd's main thread again.
Detach is useful if you have a control operation that would be too CPU-intensive and would disturb dsp.

join:
The message to the inlet will be sent to the outlet during the next clock callback of the scheduler. It doesn't matter, which thread the message was coming from. Messages from the main pd thread will be rescheduled.
Join can be used everywhere, where a message has to be rescheduled, this is also useful to place behind any threaded external calling the outlet functions from the helper thread, to make sure the following messages are being run in the main pd thread.

WARNING:
detach/join provide the possibility of threaded patching, beware of the risks:

not every pd object is thread-safe (if unsure, look at the source code or ask at the pd-list or me)
pd is not completely thread-safe itself
everything that is triggered by the detached message will be detached - so don't mix threaded and non-threaded message paths unless you know what you are doing - use the join external to synchronize with the main pd thread!
if detach/join crashes pd during a performance, don't complain ...
both detach and join have an overhead ... so only use them, if you really need threaded patching, e.g. have a subpatch that has to run for quite some time and would interrupt audio or something else

Version Control System URL:

threadlib SVN repository

Release Tarball:

threadlib source, version 0.1

Release Binary:

threadlib0.1 linux binary

threadlib0.1 windows binary

threadlib0.1 OS-X binary

adaptive

grh — Wed, 24 Jun 2009 14:20:02 +0000

Short Description:

PD-external library and patches for adaptive systems and filters

Started in:

2005

Authors:

Georg Holzmann

Authors:

Gerda Strobl

License:

GNU General Public License (GPL)

Programming language:

C, Pd

Overview:

adaptive is a Pure Data external library for adaptive systems and filters.

An adaptive system tries to learn from signals of the past. Mostly it is a simple FIR filter, whoes coefficients can be learned to model or predict an other system.
Some applications: system identification, echo cancellation, linear prediction, interference cancellation, self-tuning control, ...

For a lecture at the TU-Graz we implemented various versions of the LMS and NLMS algorithm (= least mean square and normalized least mean square adaptation algorithm).
We also added a lot of examples (see PD-patches in example folder): system identification, interference cancellation, adaptive equalization, decision-directed equalization, ...

For more documentation look through the help and example patches!

Version Control System URL:

adaptive SVN repository

Release Tarball:

adaptive source, version 0.1

Release Binary:

adaptive0.1 binary for linux

adaptive0.1 binary for windows

pix to pdp bridge

grh — Wed, 24 Jun 2009 14:14:26 +0000

Short Description:

conversion object between GEM and PDP

Started in:

2005

Authors:

Georg Holzmann

License:

GNU General Public License (GPL)

Programming language:

C, C++

Overview:

Two main video processing frameworks exist for Pure data: Gem and PDP.

pix_2pdp captures the contents of a Gem pix object and transforms it to a PDP packet (use colorspace YUV for the fastest conversion).
So it's possible to convert from Gem to PDP to Gem to PDP to ...

What can you do with it?

use pix_video/pix_2pdp for firewire input in PDP
use pix_2pdp/pdp_rec~ to record Gem pixes
maybe switch between Gem and PDP all the time
...

pix_2pdp is based on code of gem2pdp by Yves Degoyon and is a part of the gem2pdp package.
Many thanks to IOhannes M Zmoelnig for support!

Implemented colorspaces: RGB, RGBA, YUV, Grey.
For more information how to use it look at the help patch.

Version Control System URL:

SVN repository of the gem2pdp project

PDContainer

grh — Wed, 24 Jun 2009 13:53:30 +0000

Short Description:

datastructures from the C++ STL for Pure Data

Started in:

2004

Authors:

Georg Holzmann

License:

GNU General Public License (GPL)

Programming language:

C++

Overview:

::: GOAL OF THE LIBRARY :::

This library was made for algorithmic composition and of course for all other algorithms. I came into troubles with making bigger musical structures in Pure Data with send-receive pairs, arrays, etc. So I tried to make it possible, to have access to some storage in a whole patch.

::: DATASTRUCTURES :::

As storage datastructures, I implemented the C++ STL (Standard Template Library) Containers in PD. Currently following datastructures are available (prefixed with h_): h_map, h_multimap, h_set, h_multiset, h_vector, h_list, h_deque, h_queue, h_priority_queue and h_stack.

::: NAMESPACES :::

For communication I use namespaces. Every Container with the same namespace (and the same container type) has access to the same data. So you can modify and get this data everywhere in the patch. For local namespaces use names with $0.

::: DATATYPES :::

In the containers you can save all of the PD build-in datatypes: lists, floats, symbols and pointers - so it can be also used for PD's graphical datastructures.

::: SAVE/LOAD :::

All the data of all containers can be saved to disk as XML or text files. So you can also manually edit the file with an editor (which is sometimes much faster) and then load it in PD into a container. You can also load data from other containers.
Please use the XML file format if possible, because it's easier to edit in an external editor and the XML parser is much more stable.

For more documentation look through the help patches!

Version Control System URL:

PDContainer SVN repository

Release Tarball:

PDContainer source, Version 0.2

PDContainer source, Version 0.1

Release Binary:

PDContainer0.2 linux binary

PDContainer0.2 windows binary

PDContainer0.2 OS-X binary

holzilib

grh — Wed, 24 Jun 2009 13:36:59 +0000

Short Description:

collection of useful pure data abstractions

Started in:

2004

Authors:

Georg Holzmann

License:

GNU General Public License (GPL)

Programming language:

Overview:

holzilib is a collection of useful patches for Pure Data.
They were created over the years - maybe they are also useful for someone else.

Here is a (not complete) list of objects:

=========================== control ============================

binomv		binomial distribution
change_s	like pd's change object, but for symbols
clock		simple GOP clock to always have the right
		timing ;)
count		a simple modulo-n counter
datamute	mutes the data for a specific time
for_loop	for loops with pd (using [until])
mapper		maps a number between different intervalls
metronom	a graphical (GOP) metronom
randfloat	a random float between the given boundaries
randint		a random integer between the boundaries
sawtooth	message sawtooth
seq1		a message sequencer for a single voice
seq1GUI		a GOP gui for [seq1]
seq6GUI		GOP sequencer with 6 voices which can be
		synced
speedlim_any	like [speedlim], but for any types of
		messages


h_pdf		draw your own probability density function
h_pick		picks out a random element of the list
h_pickout	like h_pick without putting back the elements
h_randfloat	a random float between the given boundaries
h_randint	a random integer between the boundaries

============================ audio =============================

compr~		a full featured compressor (needs zexy external)
compress~	remove all dynamic of a sound (experimental)
delay_transp~	delay based pitch shifter
denoiser~	spectral denoiser
eq~		3 band parametric EQ (needs iemlib)
exact-delay~	sub-block delays
fad~		a volume fader with [line~]
feeddel~	feedback delay line
filtercolor~	color your sounds with a LOP and VCF combination
granulator~	a full feature granular player
in~		audio input with VU meter (needs iemlib)
invpolar~	polar to rectangular coordinates conversion
material~	formant filters of various materials
mixer4~		4 channel audio mixer
over_drive~	overdrive audio effect
od_gui~		a GOP gui for [over_drive~]
panner4~	4 channel stereo panner
panner~		1 channel stereo panner
phasevoc~	phase vocoder for pitch shifting and
		time stretching
physical~	simple physical modelling with sound input
pianorev~	spectral storage
pitch_shift_del~	another delayline based pitch shifter
polar~		rectangular to polar coordinates conversion
reverb2~	simple, cheap stereo reverberator
reverb2gui~	GOP gui to [reverb2~]
rsample~	sampler which plays random slices of one
		soundfile (using xgroove~)
sampler_loop~	looping sampler (needs xgroove external)
sampler_*	different samplers, all based on the xgroove 
		external
sine-env~	sine envelope generator
spectrum~	plots a logarithmic spectrum of a sound
vo_coder~	FFT based vocoder
volctrl~	audio output controller with VU meter
		(needs iemlib)
volctrl4~	audio output controller with VU meter
		for 4 channels (needs iemlib)
volhead~	audio output controller with VU meter
		and additional headphone output
ws~ + ws-base~	wave shaping objects

Release Tarball:

holzilib version 0.2

holzilib version 0.1

TheBrain

grh — Wed, 24 Jun 2009 12:38:57 +0000

Short Description:

TheBrain is a small C++ library for artificial neural nets

Started in:

2005

Authors:

Georg Holzmann

License:

GNU General Public License (GPL)

Programming language:

C++

Overview:

TheBrain is a small C++ library for artificial neural networks.
Currently I implemented a feedforward, a recurrent neural net and wrappers for GEM, which calculate audio signals out of video frames.

TheBrain consists of the following 2 objects:
pix_linNN (with a linear feedforward neural net) and pix_recNN (with a recurrent neural net).

pix_recNN/pix_linNN are thought as an instument/interface.
This instrument should be useful as a general experimental video interface to generate audio. You can train an artificial neural net with playing audio samples to specific video frames in real-time - so you are able to produce specific sounds to specific video frames and you can control the sound with making some movements, colors, ... (whatever) in front of the camera.
The main interest for me was not to train the net exactly to reproduce these samples, but to make experimental sounds, which are "between" all the trained samples.

pix_linNN has one neuron per audio sample: this neuron has three inputs (a RGB-signal), a weight vector for each of the inputs, a bias value and a linear output function.

pix_recNN uses a 2 layer recurrent neural net (which is much better for time-based information like video/music).

Version Control System URL:

SVN repository of pix_linNN and pix_recNN

Release Tarball:

pix_linNN source tarball

pix_recNN source tarball

Release Binary:

pix_linNN binary for linux

pix_recNN binary for linux