How to calculate MOS Scores for VoIP traffic

[karrma]

I have a need to calculate MOS Scores for my VoIP traffic. We are using Cisco CallManager. I know the codec, jitter, latency, and packet loss on each call. Is there any way to take this information and calculate a MOS Score?

[Mr. Boo Bear]

I have no idea but your question piqued my interest, so I looked up MOS:

(Mean Opinion Score) The quality of a digitized voice line. It is a subjective measurement that is derived entirely by people listening to the calls and scoring the results from 1 to 5, with a 5 meaning that speech quality is perfect. The MOS is an average of the numbers for a particular codec. Because MOS testing requires carefully prepared and controlled test conditions, the best way to get an MOS test done is to have it outsourced.

[Zachary]

You cannot calculate MOS values. You could try to locate a table that has some MOS values in with the matching numbers for jitter,loss and delay. I have build an aaplication like this some time ago and cam pretty close to actually predicting a sort of guestimated MOS value.

Since than PESQ has been introduced. Read about it at:
http://www.malden.co.uk/products/dsla/pesq.htm

PESQ can be used to measure voice quality by using programs rather than test panels.

[Big Fish]

Or you could ask your supplier, or search on the cisco web site - they must have been asked for this a lot, since it is part of the evaluation often used for commercial bids.

[Janos™]

This is the trouble with MOS. You cannot state any MOS values untill that specific network with all it's users, applications and so on is in place. Only after this it is possible to bring in a test team and start making calls.

Of course if you ask any vendor they will happily claim a MOS better or around 4 should be possible.

[Team Leader]

You can calculate a MOS score based on codec, jitter, latency and packet loss, but you need to know the "bursty-ness" of the jitter and packet loss.

For example, if you use G.711 with PLC, and have a 3% loss rate, the distribution of errors makes a big difference. One packet lost exactly every 33 packets will not noticeably affect the quality of the call. But if all the errors are grouped together you will notice a significant voice quality degradation (e.g. you may have part of the call with no loss, and part of the call has a 20% packet loss rate.)

Jitter may or may not count, what matters is how it affects the jitter buffer. A small amount of jitter may not affect the buffer at all, a little more jitter may just cause the jitter buffer to adapt by growing larger (increasing the end-to-end delay) while a larger amount of jitter may cause the jitter buffer to be flushed (resulting in a noticeable impairment).

http://www.telchemy.com/references/voice_quality.html has more detailed information. Even more detailed information can be found at http://www.telchemy.com/technicalreference.html

[ricdale]

This is the trouble with MOS. You cannot state any MOS values untill that specific network with all it's users, applications and so on is in place. Only after this it is possible to bring in a test team and start making calls using Voip Business.

Of course if you ask any vendor they will happily claim a MOS better or around 4 should be possible.

That is the problem... any and all vendors do claim that a four should be possible!
What is one to do?????

[ricdale]

You can calculate a MOS score based on codec, jitter, latency and packet loss, but you need to know the "bursty-ness" of the jitter and packet loss.

For example, if you use G.711 with PLC, and have a 3% loss rate, the distribution of errors makes a big difference. One packet lost exactly every 33 packets will not noticeably affect the quality of the voip call. But if all the errors are grouped together you will notice a significant voice quality degradation (e.g. you may have part of the call with no loss, and part of the call has a 20% packet loss rate.)

Jitter may or may not count, what matters is how it affects the jitter buffer. A small amount of jitter may not affect the buffer at all, a little more jitter may just cause the jitter buffer to adapt by growing larger (increasing the end-to-end delay) while a larger amount of jitter may cause the jitter buffer to be flushed (resulting in a noticeable impairment).

http://www.telchemy.com/references/voice_quality.html Business Voip System has more detailed information. Even more detailed information can be found at http://www.telchemy.com/technicalreference.html

I have not had any problems with jitter.. Am I just lucky or what?