Missing Folder - InvertPch?

Started by pmj, October 01, 2023, 03:52:16 PM

Previous topic - Next topic

pmj

I use the current version of SoundShaper (6.1 / Jan 2023) and noticed that if I try and run INVERT PITCH from the Spectral section of the REL8 menu that it tries to find the following map file:

c:\cdpr7\txt\InvertPch\map.txt

It doesn't look like I've ever had an InvertPch folder or that map.txt file, does anyone know which update or version it was in and where I can find it?

Robert Fraser

When the user selects a datafile, Soundshaper looks in a default location (folder) for an appropriate file and displays it if it exists. In the case of an Invert-Pitch Map, no such file has been supplied, so there is no missing folder.

One reason for not supplying a template file in this case is that the data would be rather specific to the sound you would want to invert and how you personally would want to warp the intervals. The file itself is easy enough to write: if, say, you wanted every semitone rise to go down 11 semitones, you would write 1 -11 (the data is paired). and if you also wanted every descending semitone to go up eleven semitones, you would enter: -1 11.


pmj

Thanks for explaining that, Robert.  Much appreciated.

Robert Fraser

I've posted a general follow-up called "Data Files in Soundshaper". A little more on InvertPch, though:

• There appears to be a bug when processing stereo files. The script will need revision and I'll look into that.
• The original CDP InvertPch works with extracted pitch-data files (.frq). These are then combined with a formant file (.for) for re-synthesis.
  Soundshaper's "Invert pitch contour" (not the "Alternative method") does this all within one script, extracting and altering the pitch data and then resynthesising using a formant file taken from the original sound. (The lower-level method is still available via the PitchData menu.)
• I ran a test using a simple speech file of four words, three of which are clearly around MIDI 55 (tenor G3) while number 3 dips down a semitone.
  I used a map file of -1 11 and 1 -11 and an ABOUT value of 55 (ie invert around G3) and got a fairly convincing inversion of the sound an octave higher with the semitone shift going up instead of down (G4 and Ab4).
• A different ABOUT value (60) gave a similar result transposed up to E5/F5: I'm not sure I understand the logic of the inversion here (I didn't write the program!).
• The "Alternative method" uses the Release 8 (forthcoming) SPECFNU 15. This processes formant values directly. To my ears this would be less useful - more muddy.
• Overall, playing with formant and/or pitch can only degrade the original sound quality in some way. For example, altering pitch only "works" to the extent that the new frequencies are also present in the formant file it's combined with. (For this purpose, you might consider transposing the original sound and extracting that formant file.)
• I tried extracting the source's pitch as a text file and displaying that in the Graph Editor, the idea being that the display might tell you the pitch content. It probably does, but it looked over-detailed and I trusted my ear and a nearby keyboard better!