Voice presets for sound.speech: named sets of the synthesis knobs.
Presets: narrator, assistant, computer, robot, whisper, evil,
imp.
| knob | description |
|---|---|
f0 | base pitch, Hz |
f0_range | intonation depth: 0 is monotone, 1 natural, above 1 theatrical |
speed | speech rate; 1 is normal |
formant | formant scale, the vocal tract's length: about 1.15 reads female, below 1 larger and deeper, 1.3 a small creature |
breath | 0 to 1: trades voicing for aspiration; 1 is a whisper |
flutter | slow random pitch wobble; 0 is steady |
growl | 0 to 1: alternate-period modulation, a creaky voice; 0.2 to 0.5 is useful |
quantize | hold synthesis parameters in blocks of N 5 ms frames, a stepped machine voice; 0 or 1 is off |
gain | output gain; the peak is held at 0.95 |
A knob missing from a voice takes the narrator value. Tune presets by
ear with tests/sound/manual/speak.lua.