Wingmiester, I'm pretty sure your problem is unrelated, although I don't have a solution for you - sorry (it sounds like a bug though).
The khz refers to the sampling rate/frequency which is essentially the sound quality. Since most telephone systems today run on 8khz audio, many Text To Speech developers make two versions of their 'natural' voices; a 22-44khz voice for desktop applications and an 8 khz voice for phone applications.
Normally there is no issue with using a 22khz (or higher quality) voice in IVM or any other telephony system, but the 8khz voices sometimes sound better because they are specifically designed to perform best over the phone.