Composers Desktop Project Forum

It looks like the assumed procedure (by Trevor Wishart) is to obtain the pitch breakpoint file from the (or a) source soundfile, using "Repitch getpitch 2..." (or Repitch->extract etc in Soundloom). Perhaps editied or modified using any number of tools to process breakpoint files. It is a plain time/freq format file. Of course, "repitch" requires an analysis file, which will requires running pvoc on the source soundfile first. If you are on a PC, you always have the GUI option of using BrkEdit, but I think there is a basic GUI draw/edit facility in Soundloom as well. I think a core idea is to start with repitch getpitch, and then apply changes to the breakpoint data (e.g. data reduction, shifts etc) according to the compositional goal.

How practical it is to apply such changes manually of course depends on the length of the source and the density of pitch movements - the generated breakpoint file may easily contain 100s of points. There is however no serious technical reason not to use short hand-written data files - the spirit of the software is very much that of experiment, trial and error - but the duration of the breakpoint file (final data line) should normally match or exceed that of the source file.

The FOFS tool in Soundloom uses the commandline program "psow" behind the scenes, and it may be useful to run that directly to see the usage messages, in conjunction with the general CDP documentation.

In Soundloom, the somewhat convoluted process starts with running pvoc on a source soundfile, and then selecting the analysis file as the input for "Repitch" - save the text output to some name, to ensure it appears in the Workspace (may need to click "Refresh"). Select the original soundfile, to make FOFS available in the Process window, and in the relevant parameter page use "Get File" to select the required breakpoint file.

And so on.

I have the same problem.

Hi, sounds like abrassive drone music. How is the original audio? I like it.

We've found a fault with the PC version of the program REPITCH, affecting the important GETPITCH function. The version originally distributed with CDP Release 8 (dated 15th March '23) is not working. This also affects further functions which depend on pitch extraction.

An earlier compilation, dated 2 Jan. 2023, appears to work as intended, however. You can download it here: https://www.soundshaper.net/downloads/repitch.zip
In your CDP programs folder, simply replace the 15th March '23 version of repitch.exe (if that is what you have) with the January one.

For new users, the PC program set for Release 8 will be updated tomorrow with the working version.

CDP has many processes requiring two input files. In Soundshaper these are normally loaded into two processing rows - A and B. Row A is selected by default but if necessary, click on Row A, then import a file by drag+drop or file menu, etc. into cell A_0. Then similarly click on row B and import the second file into cell B_0. Re-click Row A if this will be the primary source.

Select your process using the menus. A red box appears on the left side for selecting the 2nd source. Simply click on cell B_0. Then click OK in the box. If you are using spectral processes (for formant applications), both files should be converted automatically to spectral .ana files, behind the scenes. Select parameter values and click RENDER. The result should now be in cell A_1 - the next in the A row.

A simple example is CROSS, similar to VOCODE, which replaces the spectral envelope of the A sound with that of the B one. If the A is more static and the B rhythmic, you get a good result.

A word of caution: CDP expects the two files to be of compatible type. Processes won't work if one is 24-bit and the other 16-bit, for example.  If not sure, click on the cell and select CTRL+I for file information. All CDP spectral processes are natively mono-only, but Soundshaper can handle stereo files. I tested my CROSS example with a mono sound B modulating a stereo sound A and this worked, but I can't guarantee this for all CDP processes, unfortunately.

In Soundshaper's Patch grid, each row is a chain of processes from left to right, and the last cell in a row is the latest one for that chain, ready for the next process. When using two or more inputs, the 1st (or main) input will always be the last in its row, but the secondary ones can come from any cell, even one earlier in the same row. For example, if the source is in A_0 but the processed output in A_1 is significantly quieter, you might select LEVEL | BALANCE then click A_0 as the second input. This would balance the original and the processed sound.

Further information can be found in "Soundshaper Basic Operation", a pdf available from the "Graphic" page in the CDP online documentation, which you can access from the CDP website.