RUNNING THE CDP SYSTEM
PC / UNIX ENVIRONMENT

PC: Set CDP Path

In Windows' Control Panel, you will need to set Environment Variables, which are default values used by CDP, whether it is run from a Graphical User Interface (GUI) or the Command Line. Soundshaper sets these for you according to its program settings, but it is recommended also to set them at the PC System level.

Windows 10 and 11:

• In the Windows Search Box or Settings "Find a setting", start typing "Environment". The following should appear:
  
  
• Select Edit the system enviroment variables and then click the Environment Variables button.
• In the bottom pane "System Variables", select Path and click OK.
• In the next window "Edit Environment Variable, click New and then type in the path to _cdprogs:
  
  
  
• Click OK and again as necessary to leave Windows Settings.

Pre Windows 10:

• To set the Path for Windows 2000-8.1, go to:
      Control Panel > System > Advanced System Settings > Environment Variables
• In the UPPER pane, select the Path line and click Edit.
• Put a semi-colon (';') after the last entry (if necessary) and then add the CDP programs path, followed by a semi-colon.
  The path chould be: [drive-letter]:\cdpr8\_cdp\_cdprogs (e.g. C:\cdpr8\_cdp\_cdprogs).

PC: Setting the Memory 'buffer' size

The 'buffer' here means temporary storage space used during the operation of a program. Some programs require a great deal of buffer space and may complain that there isn't enough.

In the UPPER pane of Environment Variables , enter as NEW the variable name CDP_MEMORY_BBSIZE and the value 1024. This value may need to be expanded if performing certain operations, such as FILTER ITERATE on very large soundfiles.
In Soundshaper, it can be set and adjusted under Settings. If using SoundLoom, you will need to set it here, under the Control Panel Advanced settings.

PC: Setting a 'temporary' soundfile which can be overwritten (optional)

For command-line use, you can also set an environment variable so that you can use a specific named soundfile as a temporary file which can be overwritten.

The idea is that you can try out a program with various parameter changes without having to change the name each time, and without creating lots of different soundfiles. When you find one you want to keep, you just rename it or make a copy of it. The disadvantage of this is that you can't compare the results with a previous version, and if you have not noted what the previous parameters were, you may never be able to reconstruct the former version.

This option does not apply to to Soundshaper, which outputs temporary files in any case and erases these when they are altered.

The environment variable is CDP_OVERWRITE_FILE and its value is the name of your proposed file, e.g., test.wav. The extension (.wav) here appears to be mandatory, even if it is not used in the command line or in a batch file.

MAC & SoundLoom Installation

Installation Notes for the MAC are downloaded from the CDP Downloads Page: InstallationDocumentsPC-MAC.zip and are also found within this documentation (see links below).

MAC users should follow the detailed document CDP-InstallationNotesMAC. The distribution installs the core CDP files in the folder cdpr8 in the user's home directory.

There is also a document for manual configuration of CDP from the command line, for users of Mac OS X 10.8 (Mountain Lion) and later: Manualconfig-MAC.pdf. .

MAC: Directory Structure

The MAC directory structure is similar to that for the PC, except that the top level cdpr8 is installed inside your home directory, i.e. ~cdpr8 (where the tilde character is standard Unix shorthand for your home directory), and directory levels are separated by the / character, instead of \, as on the PC.

The structure is described in detail in the document CDP-Check_Installation-MAC, also supplied within InstallationDocumentsPC-MAC.zip.

Return to CDP Main Index
Return to top

2. CDP: BASIC CONCEPTS

Sound transformation

The CDP System is designed mainly for sound transformation, although it also has a small but growing number of synthesis functions. The emphasis is on the design of sounds, either for electroacoustic music composition or sound clips or tracks in songs or other media. Many people use CDP alongside other software in a hybrid studio environment.

CDP is a non-realtime system. It does not process an audio stream, but generally takes a pre-existing file, such as a 'sampled' soundfile, and produces another file as output. Processing is normally non-destructive: CDP will always create a new file, rather than over-write an existing one. However, in many cases, it can operate faster then realtime and accomplish complex processing in just a few seconds.

You can operate the CDP software via a graphic user interface (GUI) or via a Command line interface (CLI), depending on preference and the particular task in hand. The reference documentation describes the CDP command line(s) for each process, as these must also be written by any GUI. Command line use is greatly extended by writing batch scripts (PC) or the equivalent shell scripts (MAC/Linux).

Soundfile processing

Roughly half of CDP processes transform soundfiles, which can be of type .wav or .aif. It is now possible to use just the .wav format on both PC and MAC. Many sub-types of .wav files are supported and you can convert between them using COPYSFX, whose documentation lists the various supported types. Ambisonic B-Format files (.amb) are also supported. Sometimes an imported soundfile made by other software does not work with CDP, due to differences in the header contents. This is usually resolved with COPYSFX:
   copysfx importedsfile.wav cdpcopy.wav
– the new copy should be CDP-compatible.

Soundfile processing is in the Time Domain. A time domain display of the sound shows samples: time units and amplitude levels at that time point. The graphic display shows time on the horizontal axis and amplitude on the vertical axis, positive values above the zero line (speaker cone moving forwards) and negative below (speaker cone moving towards the back). This type of information and display is used for processing soundfiles. The image below shows a time domain signal display:

Apart from filtering, soundfile processes do not generally alter the fundamental timbre (tone-colour), but affect the evolution of the sound in time, for example by altering its loudness curve (ENVELOPE functions), changing its speed (MODIFY SPEED), extending it (EXTEND functions) by reading the file in unusual ways (e.g. ZIGZAG), echoing (e.g. EXTEND ITERATE) or otherwise repeating the whole file or sections of it (see especially TEXTURE), or layering segments to create thick grainy textures (MODIFY BRASSAGE). Spatial effects are prominent too, since the development of multi-channel techniques.

Spectral processing

CDP also has a large suite of spectral processes in the Frequency Domain, operating directly on frequency information. The input and output files here are frequency analysis files (.ana), which are converted to and from soundfiles by the Phase Vocoder program PVOC.

A frequency domain (FFT) display shows partials (frequencies) and their amplitude levels:

Partials are harmonic if they are integer multiples of a fundamental frequency. The partials are bound together, as it were, into what is perceived as a single pitch. Inharmonic partials are fractional (or decimal) multiples of the fundamental and can be perceived more individually, for example like the different tones in a bell or a gong. The makeup of the partials in a sound and how they alter in loudness over time determines the timbre of a sound.

In the image above (an analysis of an organ note), the most prominent spectral peaks show the first 12 harmonics of the sound. Above a frequency of about 4KHz, the peaks are indistinct, or randomly spaced, indicating the presence of noise. The image shows a single spectrum – either a snapshot in time or an average over time; in most sounds, the spectral content evolves throughout the lifetime of the sound.

Spectral processes often alter the frequency content, such as transposing it without altering time (REPITCH TRANSPOSE), shifting it (STRANGE SHIFT), re-tuning it (PITCH TUNE), or filtering out unwanted frequencies (HILITE FILTER, HILITE TRACE).

However, many others manipulate aspects of the spectral envelope – the overall evolution of the spectrum over time. Examples include sustaining frequency bands (FOCUS ACCU), freezing the sound (FOCUS FREEZE), blurring by time-averaging the spectrum (BLUR BLUR) and shuffling analysis time-windows (BLUR SHUFFLE). The spectral envelope can itself be extracted as a formant file (.for), and combined with pitch information to create interesting hybrids.

One offshoot of frequency analysis is the extraction of pitch (in a pitched sound) as a separate file (.frq). Various processes manipulate this information before re-combining it with a formant file to re-create sound (see REPITCH). Another possibility is to re-combine the pitch trace file (.frq) with a pitch transposition file (.trn) made with REPITCH COMBINE or REPITCH COMBINEB: see the 'Repitch Inputs & Outputs' Chart.

Analysis Bands and Windows

Two key concepts when dealing with spectral processes are Analysis Bands and Analysis Windows.

When a frequency analysis is made, the Fast-Fourier Transform (FFT) which is used converts from amplitude/time to amplitude/frequency. It does this by stepping through the source sound using a series of tiny time windows, each containing a single spectrum, and ascertaining what frequencies are present and how loud they are. This is done by dividing the frequency range into a number of bands and 'looking' in each band. These are the analysis bands, and the narrower they are, the less likely they are to have several important frequency components ('partials'). There is a trade-off to be made between band width and window-length: when more samples are used at once, the frequency resolution is better, but a larger unit of time is used, meaning a loss of time-varying data.

The PVOC analysis default values work well enough for most purposes and you can use the SPECINFO functions to find information you might need to set spectral parameter values for CHANNELS and WINDOWS.

For a more technical discussion, see the Technical Glossary, especially the articles on ANALYSIS SETTINGS and WINDOW, and The Operation of the Phase Vocoder.

Parameters and flags

Apart from the Infile and Outfile, for most CDP processes you have to set parameter values. Parameters are the controls you have over a process, corresponding to the physical dials or faders on a piece of audio hardware. On the command line, you have to type in these values, in the correct order. In a GUI, there are sliders or dials to help you set something more quickly and in any order. The image below comes from Soundshaper (process: FILTER SWEEP).

Notice that most values are numerical and that a value range is given. (You are not obliged to keep within the range, but risk a CDP Error if you are not clear about the actual range for the process.)

TIme-varying values are very common in CDP. To achieve this, the single value is replaced by a text file containing a series of time value pairs. This type of textfile is known as a breakpoint file and typically has the file extension .brk (though .txt is also permitted). In the FILTER SWEEP example, all the parameters shown can be time-varying (indicated by the 'T-V' checkbox) and in the screenshot the first has been replaced by a textfile.

In a breakpoint file, time is normally given in seconds, down to a resolution of six decimal places. There is a CDP convention that time must increase from one line to the next (even if by a minute amount). There is a further convention that where the next breakpoint value is different, the software linearly interpolates intermediate values between the user-specified values at different times, creating a ramp between the two values. To maintain a steady value, give two adjacent time-points the same value.

Some parameters are required to be textfiles, and there is a large variety of these. A key document outlining them all, with examples, is CDP Files and Codes, accessible from all reference documentation pages.

Optional parameters: many parameters are optional; in the reference documentation, these are shown within square brackets, such as [-ddelay]. Most are preceded by a flag – here, the text -d. (In a GUI, the flag will be inserted for you.)

Other optional flags: other options – to be included or not, as you wish – use a stand-alone flag (e.g. -d). (Again, a GUI should place this for you, if selected.)

Random values: in some processes, notably those in the TEXTURE suite and the granular program SAUSAGE, the value is decided randomly between lower and upper limits set by the user.

Return to CDP Main Index
Return to top

3. FURTHER KEY AUDIO TERMS

Amplitude and Gain

Amplitude refers to the loudness of a sound. Gain is the degree of change to amplitude, and can be expressed as a linear amplitude change or as a dB (decibel) change. A Gain Factor of 0.5 (i.e., where every sample value is multiplied by 0.5) corresponds to a change of -6dB. The CDP software uses mostly a 0 to 1 numerical range, or sometimes a decibel range. In the numerical range, the value 1.0 defines the peak level, while 0.0 signifies silence. In decibels, this range is 0dB (full amplitude) down to -96dB (silence). With 24bit or 32bit audio, dB values can be even lower than -96dB. CDP provides a Gain-dB chart to assist in relating these two scales.

Excessive use of gain can result in digital noise, and distortion can occur when the signal level rises above the peak level (>1.0). This can happen very easily when mixing soundfiles, as the amplitudes are summed.

Envelope

The changing amplitude of a sound over time is known as its Envelope. CDP has a large range of envelope-shaping functions as well as well as several which make use of the amplitude peaks in a sound.

The Attack Transient is the rate and amount of amplitude increase at the start of the sound – both a natural physical characteristic of many sounds and an important musical gesture. Often it is only from the attack that we recognise what is making the sound. It also affects what we perceive as the timbre or tone colour of the sound. Decay is how the sound ends, the rate at which the sound returns to silence.

Sample Rate

Sample rate is the number of samples per second with which the physical sound is represented. CD quality is 44100 samples per second. For the highest quality "pro-audio" production, higher sample rates are in common use, such as 96000. In the CDP software, output soundfiles are given the same sample rate as the input soundfile, whatever it is. There is also software (COPYSFX) to change the sample rate of a soundfile. Rates less than 44100 are not recommended.

Zero Crossings

Zero crossings occur when the waveform crosses the zero line, that is, has no amplitude. This may happen at regular or irregular time intervals depending on the complexity of the waveform. CDP refers to portions of the signal spanning two zero-crossings as wavesets or 'pseudo-wavecycles', and they are used as the basis of the DISTORT programs. Editing a soundfile at zero-crossings ensures that there will not be a click at the start of the cut portion of sound.

Return to CDP Main Index
Return to top

4. COMMAND LINE INTERFACE

CDP programs can be run from a Graphic Interface (GUI) or a Command Line Interface, known as the Command Prompt (PC) or the Terminal (MAC / Linux). All CDP processes and functions are invoked via a command line – even the graphic interfaces construct and run command lines behind the scenes. (See Command Line below for more on the structure of a typical CDP command.) The main CDP reference sections document the command-line usage for each program.

The command-line environment enables you to access CDP at the lowest level, for example to run system functions such as DIRSF or COPYSFX, to create shell scripts (MAC) or batch files (PC) which can be used to run several processes at once, or to control CDP programs in a more algorithmic way via a scripting language.

This section is mainly concerned with the PC command program CMD.exe, although most of the concepts also apply to Unix systems. For specific advice on running CDP commands on the MAC, see Using Terminal.

Current Folder

In a command-line environment, one folder (directory) is chosen as the current working directory (a kind of "you are here"). This is given in the displayed command prompt, for example:

E:\cdpr8\TEMP>

In recent versions of Windows (8 onwards), you can open a folder in Explorer and click File | Open Command Prompt: the current directory will then be that folder. In earlier versions, such as XP and Windows 7, the command window may open in (for example) a subfolder of C:\Documents and Settings\ or C:\Users, and you must change the directory to your desired folder using the cd (change directory) command. A simpler workaround is to place a copy of CMD.exe in your normal working folder, if you have one, and make a shortcut to this.

The current directory is where CDP looks by default for its input file(s) and places any output file. You can specify a file by its name alone (e.g. mysound.wav or transpose.brk). If you want to specify a file in a different folder, you must prefix the full path (e.g. E:\CDP\SNDS\mysound.wav). (For soundfiles, the suffix .wav may be omitted if you have set this as the default soundfile extension: see Setting the soundfile extension).

Path

The computer has to be able to find the CDP program you specify. This will not normally be in your working folder, so you must specify the search path. This can be done by typing PATH followed by the CDP program folder, for example:

PATH C:\cdpr8\_cdp\_cdprogs\

Once the system knows the path, you can then run a series of CDP commands. It is better, however, to add the CDP program path at the system level, as described above in Setting the Path, so that your computer can always find CDP programs. You will then not need to add a PATH command. (Note that the Path mechanism enables the computer to find only programs, not any other type of file.)

Command Line

The CDP Reference documentation gives the Usage Statement for each program, or you can type just the program name at the command prompt.

A typical CDP command line has the following structure:

program_name infile(s) outfile parameters [flags]

The Infile (input file) and the Outfile (output file) are normally the names of soundfiles or frequency-analysis files (.ana).

• In the usage statement:
  Optional parameters and flags are shown in square brackets. (The brackets are not included in the command line).
  Alternatives are separated by the '|' character, for example: [-z | -r].


• Order of items You should normally enter items in the command line in the order given in the usage statement.
  Items must be separated by a space (or a tab).
  Likewise, a space will normally be interpreted as the start of a new item. To get round this, enclose the item in double quotation marks, for example:
     "C:\My Sounds\newsound.wav".


• Mode If a program has modes, its name must be followed by a mode number, (for example: extend drunk 1).
  (Different modes commonly have different parameters.)


• Infile Some programs require more than one infile. Usually, if three or more infiles are allowed, the third and subsequent ones are optional. (This is shown in the usage statement as [infile3...] ).


• Re-running the command If you run the command line again, you must provide a different outfile name, unless you have deleted the previous outfile. CDP programs do not normally overwrite old files.


• Parameter values For each parameter, you normally provide a numerical value (within a given range) or, if a textfile is permitted, the name of the file (e.g. transpose.brk). Some parameters require a textfile rather than a numerical value.


• Optional parameters Some parameters (especially optional ones) are preceded by a flag, consisting of a minus sign followed by single letter (e.g. -c). The letter is normally case-sensitive and the parameter value follows immediately without a space (e.g. -c25).


• Flags Other flags indicate options which are set by typing the flag (e.g. -k ). (Omit the flag if you do not want that option.)

An example of a usage statement followed by a possible command line for the program:

grain repitch mode infile outfile transpfile [-blen] [-lgate] [-hminhole] [-twinsize] [-x]
grain repitch 1 grnsnd.wav grnpch.wav pchlist.txt -b1.2 -x

For further information on CDP command lines, see the Technical Glossary.

CDP Error and Warning Messages

If a command line has been wrongly constructed, or the CDP program cannot run the command for any reason, an Error Message will be shown. The most common causes are invalid parameter values or inability to write to the output file.

There are also Warning Messages and general advice, particularly for Information functions.

Basic Command Window operations

Users new to the command line environment may find the following basic procedures helpful:

• Select a drive: enter the drive letter followed by a colon and press Enter, for example:
    d:<Enter>


• Make a directory (folder): md directoryname <Enter>. This creates a new directory.


• Change directory: cd directoryname<Enter> if nested (i.e. contained) directly within the directory where you are at the time.
  For directories nested deeper, use, for example, cd directoryname\directoryname


• Move up a level – out of a nested directory back to the directory which contains it : cd ..


• get back to top level: cd \


• the extension for a PC batch file is .bat and for a MAC/Linux shell file, it is .sh.


• Move back to a previous command: use the up-arrow (i.e., to retrace your steps in the 'command history')


• Deleting a soundfile: when you delete a soundfile in your current directory, you must include its extension; e.g. del soundfile.wav


• Re-direction: the re-direction operator > can be used to send CDP console output to a text file, for example:
    soundinfo props mysound >sndinfo.txt
  Textfiles written in this way are normally overwritten if the filename is re-used.


• Wildcards: the 'wildcard' operator (* = "any") selects a number of similar files – use with care!
  For example:
    erase *.wav
  will delete all .wav soundfiles in the current directory.

You can read more about the Windows command line environment in a separate document: The Windows Command Prompt. This is a basic introduction, explaining directories and paths, how to change and create directories, set a path etc.

For a comprehensive description of all CMD (Windows PC) and BASH/OS X (Linux / MAC) commands, see the website ss64.com.

Command history

Once a separate utility, DOSKEY is a facility built into CMD.exe (although there is still a DOSKEY command to change its operation). Holding down the function key F1 recalls the last-used command line character by character; hence you can alter some of the content as you go. This is useful if you need to re-run a command with slight variations. You can also scroll up and down the command window using up/down arrow keys to see the whole command sequence.

Although the standard controls for COPY and PASTE are not available in the Command Window, equivalents are available in the Edit Menu (below the CMD icon): select MARK and then drag the mouse to highlight the required text, followed by RETURN to COPY it; then reposition the cursor and select PASTE as required.

Batch Scripts

Batch scripts contain one or more CDP command lines and/or general system commands, saved as a batch file. A batch file is a text file with the extension .bat. The Unix equivalent (with a similar syntax) is a shell script, suffix .sh.

To run the batch file, enter the name of the file, optionally preceded by call. You may also omit the suffix .bat. If you do, be careful not to give the batch file the same name as a CDP program or a CMD reserved word.

Batch files can speed up the use of the CDP system in a command line environment considerably. For example, if you are running a program over and over, changing the parameters a little each time in an effort to get just the right sound, putting the command line into a batch file could be useful, especially if you want to try a number of options all at once and then compare the aural results. It is also useful when working with spectral processes, where analysis and re-synthesis with PVOC are included in the batch file. Batch files also provide a record of the final versions of the command line(s) used to create a certain sound.

Here are some examples:

1. Complicated command line: a long command line contained in a batch file can be edited in a text editor.

Name: sauce.bat
Contents of file:
  modify sausage insound sausout 0.2 1.2 gsize.brk pshift.brk space.brk
(insound.wav and sausout.wav are the Infile and Outfile respectively.)
To run the file, enter: sauce. The batch file runs SAUSAGE with the flags and parameters you've specified.

2. Varied parameter values: several outputs, each with a different number of retained channels.

Name: traces.bat
Contents of file:
  pvoc anal 1 insound insound.ana
  hilite trace 1 insound.ana outsnd1.ana 20
  hilite trace 1 insound.ana outsnd2.ana 10
  hilite trace 1 insound.ana outsnd3.ana 5
  pvoc synth outsnd1.ana outsnd1
  pvoc synth outsnd2.ana outsnd2
  pvoc synth outsnd3.ana outsnd3
To run the file, enter: traces.

3. Use of parameters: a parameter value is passed to the script when run.
It can be very useful to pass a parameter, especially the name of a file, to the script. Batch scripts refer to the parameters using the tokens %1 to %9. The first item passed is always %1, the second item is always %2, and so on. The scripts use simple substitution of the text passed.

Name: p.bat
Contents of file:
  paplay -i %1.wav %2 %3
To run the file, enter, for example: p mysound 0 2. %1 is the file to be played, less the extension .wav, which is added in the script; %2 is the from parameter value; %3 is the to value. (The -i flag means "play immediately".)
On another occasion, you might enter p outsnd 0.5 ; the script cannot use the "to" value, because it isn't there, and it can be omitted in the PAPLAY command, causing the sound to play to the end.

4. Nested batch files: one batch file calls another.
sauce.bat is expanded to play the outfile if it has been created.
Parameters %1 and %2 are the infile and outfile, respectively.

Name: sauce.bat
Contents of file:
  modify sausage %1 %2 0.2 1.2 gsize.brk pshift.brk space.brk
  if exist %2.wav p %2
To run the file, enter, for example: sauce insound sausout.

Limitations of batch processing

While Batch files enable a specific sequence of CDP commands to be tested and run in one go, there are some limitations which affect more general use:

• CDP processes must be matched with suitable inputs. Most of the 'Groucho' time-domain commands can handle stereo input (a notable exception being the DISTORT suite) and many, but by no means all, can accept multi-channel input; but equally there are quite a few multi-channel commands which take a mono input and produce multi-channel output. The whole of the spectral suite and its derivatives are natively mono-only. It is very important to remember this when using the command line or batch files.
  


• Timing issues: many CDP processes are time-sensitive. For example, any time-varying data file or a function like CUT must not refer to a time that is beyond the end of the file. Hence the infile must be of a suitable length to avoid a CDP error.


• Batch parameters. There is an upper limit of 9 parameters that can be passed to a batch script: only parameters %1 to %9 can be referenced by number within the script. (While the SHIFT command provides a way of accessing additional argument values, it is not easy to script.) This limits the scope for passing CDP parameters, since, in addition to Infile and Outfile, many CDP command-lines have more than seven parameters.


• The batch scripting language is quite limited. For example, it lacks functions, which execute a block of code and can return a value as a result. (It would not be easy to read and respond to soundfile length, level and channel data, even though CDP functions can provide this information.) BASH scripts (MAC/Linux) are more versatile, but those wanting to develop more algorithmically intricate processes should turn to more powerful programming languages, such as Tabula Vigilans, Python, TCL/Tk etc.


• CDP Errors. If anything goes wrong, subsequent processes and commands tend to fail, because a particular file has not been written.

Return to CDP Main Index
Return to top