No Snips audio output via Matrix Voice ESP32

The Matrix Voice accept audio sample rate at 44100 in stereo only.
We get no readable voice from Snips apps to Matrix Voice (ESP32).

How is it possible to configure Snips audio replays at this sample rate?

Our Snips is running in an Ubuntu VM on a FreeNAS server.
We therefore have no local audio driver installed.

I have solved this issue. The problem was that I gave the MATRIX the wrong major device number.