a talk box is an audio pedal that has a compression driver (speaker) connected to a small tube that goes into the musicians mouth
- they play whatever it is they want to affect (guitar, keyboard sound, etc) and then use a microphone to pickup up the sound being manipulated by the formant structure of the player saying words
- as we say words the shape of our resonant cavities are changed to produce the characteristic sounds of syllables
- these are used in similar fashion to a filter with the added luxury of having the players lips and teeth
control the extent of the effect and the resulting talking
A vocoder on the other hand is one of two different devices
- both versions use a carrier and a modulator to create a new sound
-
the original used multiple sine waves of every possible frequency (usually in 1/3 octave bands, but more recently aimed at being every note from 20Hz to 20kHz) as the carrier and then used the modulation input of a an individual speaking to or a programmed range of speech segments to create synthetic speech - a precursor to fred and similar speech synthesis models
- the more recent version uses two wave forms: the carrier being any sound and the modulator is analysed into many bands (32-512 or even 1024) and then applied to the carriers frequency and envelope to change the characteristics of
the original sound
As for part 2 of your question, I do not know of thsi zap and roger of which you speak, so a link may be appropriate before I offer a comment on the applicability of either method