~ A Survey of the Outlook and Objectives of the CDP Founders ~

Historical Context: a very condensed synopsis

The presence of sounds and references to sound have been a part of music from time immemorial. However, sound took a distinct step towards a new kind of role thanks to the Italian Futurists. They embraced the sounds of the mechanical industrial world in which they were living. They were excited by its energy and visceral qualities. They invented noise-making machines and conceived of concrete poetry made from sounds. Things were never to be the same again. The French composer Edgard Varèse devised a 'constellations' approach to structuring sound, resulting in astonishingly original and powerful music.

Matters started to evolve more quickly when tape recorders and other equipment in radio studios began to be used (Varèse, Stockhausen, Berio ...), and Max Matthews rejigged telephone technology to produce sound in a way that would be useful for composers, creating the first sound synthesis software, leading to Music-5, Music-11, Csound and the CARL synthesis program. Meanwhile, in France, Pierre Schaeffer pioneered using recorded sound samples (known as musique concrète) and did a great deal of work on categorising sounds, work taken up by the Groupe de Résearch Musicale which produced some key software programs with which to manipulate sound samples, such as transposition and filtering.

When CDP came on the scene in 1986, therefore, much innovation had already taken place, and hundreds if not thousands of pieces had already been made from sounds, and the CDP Founders themselves had also done so using classic tape technology. What was different after 1981 was the emergence of the personal computer. For the CDP Founders, the future direction was clear, and they set about harnessing computer power to assist in composing with sounds, currently referred to as electroacoustic music.

CDP emerged from several years of discussions as the INTERFACE group, which gradually included experimenting with creating software for the first personal computers (Sinclair, Commodore, BBC Micro, Archimedes, Sharp, Atari ST). The Project took shape when all agreed to settle on the Atari ST because of its MIDI and games ports. These Founders and their initial contributions were: Dave Malham (SoundSTreamer hardware design), Martin Atkins' UNIXTM-like soundfile operating system, Andrew Bentley (some basic sound manipulation programs modelled on the GRM Tools), Trevor Wishart & Richard Orton et al. (adaptation of the CARL software and in particular the Dodson Phase Vocoder to run on the Atari ST and other program ideas), Rajmil Fischman (a graphic user interface and graphics library for the Atari ST), Archer (Tom) Endrich (contributions to system design, documentation and dissemination of the new software system).

Goals of the CDP Founders

The CDP software was developed by composers for composers. The Founders of CDP had a vision of how sound would continue to evolve as an important component of music. They wanted to use sounds in their compositions, to integrate them into the fabric of their musical creations. Their approach grew out of the historical context, but involved a number of key decisions that defined they way they wanted to work.

The result of these decisions was a system that became a powerful tool for building and organising sounds, with hundreds of functions with which to sculpt sound. As time went on, Trevor Wishart wrote more and more of the software to assist with his own compositions, so the CDP System can now be justly described as the 'CDP-Wishart Libraries', and some of the character of the software is due to his personal compositional preferences. There were accepted lacunae in the CDP software. Some were filled by commercial software and some by 'synthesis engines', but there are also a number of lacunae which CDP users still hope to fill. See Wishlist below. Most CDP users work in a 'hybrid studio', that is they make use of many other tools besides the CDP software.

First Decision: focus on musique concrète

The experience of the CDP Founders had been with musique concrète as handled in the classic tape studio. This gave them a keen appreciation for the potential of working with recorded sounds, but also made them very aware of the degradation in sound quality that resulted from repeated recording and re-recording. When sound could be digitised via computer technology (Analog to Digital converters) and promised re-recording with minimal degradation, they took to it like ducks to water. As mentioned above, they were able to build in particular on the work of Varèse, Schaeffer, and the GRM. While respecting the importance of sound synthesis, they mainly worked with digitised sounds as their preferred source material.

Second Decision: leave sound synthesis to others

The feasibility study begun in 1986 was to port the CARL software to the Atari ST, including the all-important Dodson Phase Vocoder (for FFT analysis and resynthesis). When this was achieved, the hardware, operating system and first sound manipulation tools were created, CDP had a working digital direct to disk sound transformation system (1987). This work, the complexity of this task, and its limitations as sonic material: Music-11, CARL, Csound led them to realise that their preference was to work with sound samples (sometimes referred to as subtractive synthesis). They therefore decided to leave the sound synthesis to others. The CDP software therefore has only minimal synthesis tools, mostly for diagnostic purposes. CDP however maintained a close connection with Csound, ported it to the Atari ST, distributed it with its software (under licence), and used it when sound synthesis seemed appropriate.

Third Decision: make use of the 'command line'

The UNIXTM-like nature of the original soundfiling system for the Atari ST meant that it was run from a command line interpreter. This parsed and ran typed user-instructions. This contribution from Martin Atkins was crucial to the whole nature of the CDP system and continues to the present day, thanks to the presence of DOS on the PC and the Terminal on the MAC. Graphics built as front-ends operated in a graphic manner on the surface, but behind the scenes were constructing command lines that ran the CDP sound manipulation software. The current advanced GUIs, Sound Loom (MAC & PC) and Soundshaper (PC only), still operate in this manner.

On the face of it, using the command line is a very old-fashioned and apparently cumbersome way of working. The two GUIs written for the CDP System provide options and can provide a comfortable way of working. However, there are several benefits to using the command line directly. These include:

  • speed: When command history is used, an up-arrow takes you back to previous commands, ready to be tweaked and re-run; it is in fact a quick way of working.
  • batching: Batch files can be created to run specific programs, especially if they have quite a few parameters, or even to run whole sequences of commands (sound design templates); typing is therefore kept to a minimum.
  • research: The CDP software can easily be employed in a research environment that is also using a command line environment. This is especially the case now that CDP has been ported to Linux. One area that is being explored at the moment is data sonification. The command line basis provides a working environment, and the CDP processing functions provide ways to make the sonifications more aurally effective or even musical.
  • algorithmic music: As knowledge of programming in languages such as Python, Java, TCL/TK or CDP's Tabula Vigilans ('T-V') continues to become more widespread, it is proving to be very useful for the CDP software to be able to be run in a command environment; such scripting makes it possible to run the CDP software within the context of design algorithms of an indefinitely complex nature, and to output algorithmically derived text files that are required by various CDP sound manipulation programs.
  • mobile applications: For all the user-friendliness of mobile platforms, behind the scenes they are based on scripting; this makes the CDP software an ideal platform when sound processing is required.
  • porting: The command-line basis of the CDP software makes it much easier to port the bulk of the CDP sound processing software to new platforms and operating systems, and only minimal changes have been required even for the GUIs.
  • a mobile CDP: Because of the command-line basis of the CDP software, it is feasible for it to be ported to mobile devices, given suitable processing power and screen size.
  • disabilities: The text-based command line interpreter offers an option for users with visual impairment, though only to a limited degree due to the complex command lines of the more advanced programs. (The Soundshaper GUI for PC has a considerable range of facilities to help it work with screen readers.)

Fourth Decision: focus on listening rather than graphic representation

The CDP Founders had a healthy respect for the ear of the composer. Digital sound is normally sampled at 44.1K per second, 44K for CD, and even higher resolutions such as 96K or 192K per second. In contrast, the eye is fooled into seeing continuous movement at only 24 frames per second. The limited number of pixels across a computer screen means that the visual representation of sound is highly compromised. An illustration of the difference between visual and audio is how much more difficult it is to achieve seamless morphing in audio than it is in visual.

The CDP Founders, while providing some tools of a graphic nature, have maintained a focus on the ear, on listening to sounds created. The process of sound design is very physical: imagine, create, listen. Even with the Soundshaper GUI, editing the sound is done by ear: the user listens, watching the time-counter, sets time-points for cutting, listens again to see how well they worked, and resets them as required. The ear knows when it is right. Writers of software for CDP have therefore focused on functionality rather than graphical finesse, which can be very time-consuming. The lack of graphics has cost the CDP on a commercial level, but it has freed up the software writers to work on functionality.

Besides the two GUIs, CDP also has 6 graphical programs:

  • VIEWSF by Richard Orton. This displays a soundfile waveform in the time-domain, accurate to the single sample, a feature not often encountered. You can zoom in and out and place marks for cutting. VIEWSF will also display the spectrum if given an analysis file (.ana) as input. It can be accessed from with Soundshaper. (Sound Loom also has a spectral display.)
  • GRAINMILL by Richard Dobson. Based on Trevor Wishart's earlier BRASSAGE program, this is a handy tool for creating complex granular textures.
  • BRKEDIT by Richard Dobson. This displays one or more breakpoint files at the same time, with data reduction facilities. It is useful for creating breakpoint files graphically or comparing them for diagnostic reasons.
  • DRAWSCREEN by Richard Orton. This is part of ProcessPack, a companion set of sound design modules with a graphic front end in TCL/Tk. Drawscreen makes it possible to draw different types of lines connected to the same or different waveform generators on the same screen. This is another way to make complex sounds, though limited by the types of waveform generator.
  • AL-ERWIN by Rajmil Fischman. This is a hybrid mixing environment plus a granular program based on ... 'Hybrid' means that it combines feature of sound-building and sequencing.
  • ProcessPack by Richard Orton & Archer Endrich. The aim here is to provide higher level composition modules, such as DISPERSAL, which can scatter or gather soundfiles in time and space.

While there is plenty of room for graphic development of parts of the CDP software set, CDP likes to keep its emphasis on the ear as central in the process of composition.

Fifth Decision: maximum functionality

This is what it is all about for CDP: thinking up as many ways as possible to get at and modify the salient components of a sound – software for composers designed by composers. Having said that, there are many kinds of music, and the CDP software has been designed with a bias towards modern electroacoustic music as evolved within the Western Classical tradition. The key words are variation, integration and form.

A useful parallel is the use of figure in classical music. Like the leaves of a tree which are the same but not identical, a musical figure undergoes endless variation as a composition evolves. A good part of the 'organic' nature of classical music is based on this use of figure, whether in melodic, harmonic or textural constructions. Similarly, composers working with sound create many variants of the original source material and weave them together to form a composition. It is often important to disguise the physical origin of the sound so that it can enter into the more abstract flow of the form. Contrast is itself based on being able to create the similar.

The overview of CDP sound processing given here is rooted in the need to vary, to disguise, to contrast, and to use these abilities to create form. This approach presumes a sense of parameter space and time-varying movement through this space. The parameters are getatable features of a sound, and the time-varying movement provides suppleness in the result: different values at different times and the possibility of interpolating between these values (gradual change). CDP also optimises vertical mixing. This means layering sounds together to make new sounds rather than focusing on the sequence of sounds in time, though this can also be done. Rather than thinking in terms of layers of tracks, CDP visualises layered components within the same track.

The CDP software, after 30 years of development, now offers an extensive range of facilities. The CDP Desk Reference offers the quickest overview of the software, and for more detail there are the indices for the HTML documentation and Tutorial Workshops.

Here is an overview of the areas in which the CDP software is strong and relatively unique:

  • segmentation – break up a sound into pieces (in sequence or randomised), rearrange those pieces, build extended flows, disguise the origin
  • distortion – create strong, visceral sounds by working with irregularly spaced (and therefore inharmonic) waveccyles (looks at where the wave crosses the zero point), disguise the origin
  • envelopes – extract both amplitude and spectral envelopes, apply a variety of processes to them, impose or replace on other sounds; a facility for graphical envelope alteration is provided in Soundshaper, including drawing your own envelope on a blank page automatically set to the timespan of the source sound open at the time
  • multi-event textures – multiply a sound in vertical space and horizontal time, with the option (among many others) to map the iterations to a harmonic grid or to weave in defined melodic and/or rhythmic motifs, allowing some powerful crossovers with 'conventional' pitched music
  • transitions between sounds – move between sounds whether as a direct crossfade or as a morphing of one into the other, transfer the attributes of one sound, such as envelope or formants, onto another, interleave sounds, generate a series of sounds in-between two input sounds, combine maximum amplitude value in each channel in each window from comparing two sounds
  • handling vocal sounds – based on formants analysis, these tools help to integrate vocal sounds into a more complex sonic texture, work with pitch-synchroised vocal grains ('PSOW'), impose vowels on a sound
  • pitch profiles – extract a pitch profile and work with it, such as combinations with transposition files; these introduce a linear dimension derived from contours in sounds rather than from discrete pitches
  • blurring – play with analysis windows: average, randomise amplitudes & frequences, scatter, shuffle, weave

Some other not so unique but handy facilities are: tuning in the spectral domain, time-varying, sweeping and iterative filters (some with tunable bands), facilities to make and manipulate mix files, options to make, manipulate and diffuse multi-channel soundfiles, various editing facilities. There is also a range of functions to provide information about sound and analysis files as well as utilities to handle/convert the various types of files used by the system. The complete package – now a free download! – provides a powerful and flexible sound design toolkit.

Wishlist

There are nevertheless a number of facilities that could enhance the CDP package. Our wishlist therefore includes:

  • more synthesis options – there may be some in the wings that could be brought into the CDP System with a little help to test and document them
  • blocking out and a moving play cursor in VIEWSF
  • more extensive spectrum display options, with labelling
  • a better way to isolate and work with partials in focused pitches
  • more ways to work with pitch in a conventional manner, to help with integrating (discrete) pitches and (continuous) sounds, including ways to build strong, clearly pitched sounds
  • more extensive facilities for defining and playing back rhythms, including meters, acceleration, multiple meters (simultaneous or following one another), rhythmic seives ...
  • a wider range of algorithmic facilities and their integration into the sound processing software
  • development of sonification procedures within the main body of the software


Last updated: 09 March 2016

© 2016 - CDP & HITHER GATE MUSIC, Plymouth, England - Archer Endrich