Hi. As always, if you want more quality, you have to sacrifice resources.
If I were you, I would use Voice Transport rather then Voice Networking, but if Voice Networking is used by network design and there are many nodes in the network, you will not be able to change it. Next thing, check if "silence suppression" on both Passport's is set to "off". This should help with "intermittency". But not always. The sacrifice in this case will be that each vs will hold it's bandwidth even in "no voice" condition. But switching between "voice" and "no voice" will be much smoother. To improve the quality of music, you can test different types of music (not really effective) or go to different voice transport protocol, e.g. instead of G729 with 8K channels use G728 or G726 (16K or 32K each channel). But this will decrease number of channels twice or 4 times which is HUGE sacrifice.... If something better comes to my mind, I will draw a line... Thanks... Alex.