Hey @DallasSteve – I’m a big OVox fan and a fan of vocoding in general.
There are two critical aspects to vocoding: the carrier (source voice, etc.) and the modulator (the synth.)
To get a good vocode without artifacts you want the voice signal to be as clean as possible. Any artifacts like breath sounds, clicks, pops — even if they are quiet, they can result in weird sounds once modulated by the synth sound.
Adjustments to the carrier and adjustments to the modulator can make a huge difference to the resulting sound. With that in mind, try using a good channel strip like Scheps Omni Channel or SSL EV2 on the vocal before it is vocoded. Use the gate (or manual editing) to eliminate all sounds other than speech.
Now try adjusting the EQ – try cuts and boosts on various ranges and listen to the resulting vocoded sound. Also adjust the amount of compression on the voice. It will affect the vocoded result as well.
EQing the modulator (synth sound) is just as important, but I haven’t set up OVox in a way where that is possible yet and I’m not sure if you can. I just know, from the past, that using a channel strip on both the source synth and the voice and tuning them accordingly makes a big difference with any vocoder.
Lastly, post-processing your vocoded result is just as important! So in a perfect world, you have a fully featured channel strip on the voice, the synth, AND the vocoded result.
PS. Based on your needs, do a search for “Emvoice” – it might be right up your alley. Imagine a realistic voice (male, female, or robotic) singing on your song. You enter the words and enter the notes, and it just works. It’s pretty incredible. I use it to write placeholder parts which my female singer then replaces — but it actually sounds surprisingly good just on its own.