Authors - Peter Brinkmann, Peter Kirn, Richard Lawler, Chris McCormick, Martin Roth, Hans-Christoph Steiner

Pure Data can be relevant to sound designers, musicians, and software developers, given a way to integrate graphical sound development with their tools and platforms of choice.

We present libpd, a thin wrapper that turns Pure Data into an embeddable audio library, as one solution. libpd emerged as a byproduct of an effort to port Pd to Android, but it has since taken on a life of its own, with language bindings for Java, Processing, Objective-C, and Python, mobile apps for Android and iOS, and inclusion in packages such as openFrameworks and jReality. Since its announcement in October 2010, it has attracted more than 100 developers, and libpd-based applications are running on more than 3,000,000 iPhones and available to millions of Android devices in the Android Market. We explain the design of libpd, how to use the library, and how it fits into the lifecycle of an audio application from sound design to deployment. We conclude with a discussion of the relationship of libpd to Pd proper.

Design

libpd is defined not only by what it adds to Pd, but what it removes. As a pure audio library, it has no graphical user interface, no audio drivers, and no MIDI drivers. Most crucially, libpd has no innate sense of time.

Instead, libpd focuses on signal processing in its purest form: Samples go in, magic happens, samples come out. The heart of libpd is an audio callback that accepts a buffer of input samples, processes them, and fills a buffer of output samples. libpd keeps track of time only in terms of the number of samples requested so far (i.e., the number of invocations of the process function); calling the process function at the right time is the responsibility of the client code.

The libpd API also includes a number of functions for sending messages to Pd, as well as hooks for registering callbacks that receive messages from Pd. From the point of view of Pd, this mechanism is no different from exchanging messages with [send] and [receive].

Workflow

Pd has been used in prototyping, using OSC or [netsend]/[netreceive] to exchange messages with the outside world. With libpd, the relationship between client code and audio engine is more immediate. Instead of having to launch Pd separately, Pd is embedded in the client code. Instead of using the network, Pd and the client code exchange messages directly. This setup allows sound designers and software developers to work independently; they only have to agree on basic parameters such as the number of audio channels and the labels of send and receive symbols. Software developers can treat libpd as a black box, and sound designers can build patches as usual, designing and testing them with the usual GUI elements. In order to deploy a patch, the sound designer only needs to assign the appropriate send and receive symbols to the GUI elements. The prototype is the production code.

uni-weimar.de/medien/wiki/PDCON:Conference/Embedding_Pure_Data_with_libpd:_Design_and_Workflow

AUDIENCE for Pure Data is a library for audio immersion, sound scene production and auralization. The software enables to work independently the sound scene composition, its acoustic rendering, encoding it in a suitable multichannel format, and finally auralizing it using 2D or 3D sound fields. A major feature of it is the sound scene conception independence from the listening mode, as the user can work out a desired sound scene spatial experience regardless of the loudspeaker configuration used to listen. The AUDIENCE library distribution is organized in 12 directories containing the main spatial processing layers (L1, L2, L3, L4), an auxiliary functions layer, reference applications, supportive libraries, binaries, images and sounds, test patches and documentation. Externals are compiled to MAC OS, Linux and Win32. A typical project flow involves creating a sound scene (layer 1 object), configuring an acoustic rendering scheme (layer 2 object), encoding the spatial sound of the rendered scene (layer L3 object) and then decoding and auralizing the sound scene through a loudspeaker rig (layer 4 object). Auxiliary functions, such as transport, mixing and volume controls are available from objects in the auxiliary layer. The full distribution version 2.0.2 supports a scalable interactive sound scene graphical user interface, Ambisonics (up to the 3rd order) and MPEG-4 AAC codecs, image-source-based acoustic simulation, multichannel recording and playing (including 5.1) and it can decode to 10 different speaker layouts. Currently a new acoustic simulator with ray-tracing and radiosity support, loudness management and object-oriented timeline features are under development. The architecture is scalable: many scenes and sound sources can be combined, while real time interaction placing and moving objects in the scenes is possible. The final sound field can be exported and played back in stereo and surround output formats. AUDIENCE and its open version OpenAUDIENCE are been developed collaboratively since 2005 [1], and have been used in several scientific and artistic projects since then. In 2007 it was used to spatialize the piece “The Unanswered Question” by Charles Ives (1906) with 9 instruments (a string quartet, a wind quartet and a trumpet). In the same year its underlying architecture was proposed as framework for the new MPEG SAOC[2] codec. In 2009 the electro-acoustic piece “Kitchen” by Fernando-Lopez Lezcano, with 8 sound sources, was spatialized during the SBCM 2009 symposium in Recife. AUDIENCE is presently employed as auralization engine in a CAVE system at the University of São Paulo, and has been further used as reference audio engine for reconfigurable digital music applications, soundscape design applications and scene-oriented mixing. The system development is going to its 3rd. generation, designing improvements in the usability and sound fidelity, expanding the number of spatial processing functions, and looking forward to enlarging its platform availability to prospecting new applications and user groups.

uni-weimar.de/medien/wiki/PDCON:Conference/AUDIENCE_for_Pd,_a_scene-oriented_library_for_spatial_audio

Instant Decomposer is intended for solo-performance and for collaborations in multiple-discipline projects. Think of sonic sketches, live soundtrack for experimental film or theatre production, live music for dance production, sonic support and voice-over for puppet show or animation, interactive sound installations, and so on.

uni-weimar.de/medien/wiki/PDCON:Concerts/Katja_Vetter

This work is a quasi-improvisational performance for Peacock, an self-designed box-shaped sensor-based interface. This interface detects the movements of a performer's hands employing thirty-five infrared sensors. The output from the sensors is mapped to more than three-hundred parameters of a synthesizer, running on a Pd-extended patch. The mapping between data from the sensors and the parameters of the synthesizer gradually varies as the piece unfolds. The synthesizer generates electronic sound, employing the phase-bash synthesis technique based on six seconds of prerecorded spoken voice. Thus, the sound produced by the synthesizer is always readily associable to the sonority of the human voice, though often highly distorted or modulated.

Moreover, the synthesizer allows the performer to control a maximum of five voices simultaneously and enables him/her to create flexible homophonic and polyphonic textures.n order to achieve a more organic sonority, a number of chaos attractors were developed as external objects in Pd-extended by the author, and these attractors are utilized for randomizing several parameters of the synthesizer in various sections of the work.

uni-weimar.de/medien/wiki/PDCON:Concerts/Chikashi_Miyama

Reverb Design is about creating an aesthetic appealing spacial impression aimed at sweetening the listening experience for a given context. However, the ambiguity of auditory and visual cues in reproduced sound makes it necessary to create an auditory illusion. To make this illusion work, an aesthetic concept of simplicity and intelligibility is proposed based on psychoacoustic facts and recording practice. A development kit for Pd is presented to encourage reverb design from a “bird's eye view”, with complete reverberation algorithms as basic building blocks. An example topology complements previous work with diffuse reflections and late reverberation.