[sdiy] vocoder filters

[Bernard Arthur Hutchins, Jr]

Of courses vocoders have “dynamic poles” (not fixed resonances) – else what is the point?  In the original application (Voice-Coder) the idea was to exploit the redundancy in human speech (10:1 to 50:1 - - said redundancy being the repetition of many very similar “cycles” in time that constitute a particular vowel/phoneme/syllable). Instead as few as three bandpass resonances might well represent a “frame” of 1/5 a second.  A loss of quality of stored/transmitted speech for a vast reduction of memory required (expensive in the early days). But clearly the resonant frequencies changed with each new frame encoded.

If applied to music, probably one thinks of synthesizing singing or some DESIRABLE vocal effect.  I tried this:

http://electronotes.netfirms.com/AES4.PDF

The “animator” worked through a series of (LFO controlled – see Fig. 7) of POSSIBLE formant frequencies (phonemes both used and unused in a particular language).  It worked - it sounded human!  BUT basically like a human suffering from food poisoning!   Lesson learned.

-Bernie


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20190901/a598c032/attachment.htm>