I suspect the thing you could do, is do it yourself if you have plenty of time and a multitrack of some sort. You need top record some people with different voices, doing the phrases and noises you suggested. get maybe 4 female and 4 males, anyone will do. Record each one with them doing a low voice, mid voice, high voice, loose throat, tight throat – basically as many styles as they can do. You then start to combine them. If your editor lets you pitch shift and stretch, do very small shifts, no more than a tone, probably less and subtle stretches and shortenings. You can easily get to 30-40 different voices. Pan them, a touch of reverb and it sounds great. I started doing this kind of thing in the seventies when we did a series of live audio CDs from pubs – the real audience tracks sounded rubbish, so we made a few of our own and added them, eventually using just our ‘pub noises’. You also need to record background – people make a lot of noise even when they are not speaking.