Greetings,
Here's a short follow-up to yesterday's presentation of Zebbra guys.
What Markus and Roland have presented, is an approach to monitoring the quality inside the box (CUCM in their case). But there are actually ways to monitor the quality outside of the box, and they have been around for a while.
For example, I've seen a demonstration of Malden MultiDSLA tool over 10 years ago. At that time they only had the analog interface that emulated a handset, and two Malden boxes were sending an audio sample to each other and assessing the quality of received audio. The tool is quite expensive though.
There are currently several commercial solutions for audio quality analysis:
POLQA (expensive), recommended by ITU-T: it compares two audio files and produces the quality assessment score.
Sevana AQuA (less expensive): it also compares two audio files and produces several quality metrics. I compared it with PESQ (predecessor of POLQA), and AQuA works significantly faster. It was very helpful in detecting lost RTP packets in the tests where I could not place a packet sniffer wherever I wanted.
There are also various approaches to passive quality analysis, and some are documented in ITU-T P.563 and G.107. Sevana is also offering its PVQA tool that tries to detect packet loss and audio distortions by analyzing the input audio.
I could not find any open-source tools which would help in quality analysis at the audio level. But there's plain old tshark which produces the loss and jitter statistics for RTP streams, and it's pretty efficient if you are able to capture the traffic at the receiving end. Also VoIP monitor provides the ability of G.107 analysis on captured data packets.
Here's my detailed article on using Sevana AQuA: https://txlab.wordpress.com/2015/06/02/quality-assurance-for-voip-calls-2/
And here's my demo lab, and you have a possibility to send an audio recording for Sevana PVQA analysis: http://voxserv.ch/demolab.html
cheers,