[Asterisk-Dev] Measuring MOS Scores

[Steve Underwood]

Hi Paul,

There is a lot of bogus junk like this around at the moment. That page 
actually tells you what it is measuring, and that is not MOS. In most 
places is seems carefully vague, throwin the term in brackets, and not 
clearly saying what they measure. Where they do finally say, it the 
stability of jitter, packet loss and other issues they are measuring. 
They don't even seem to go as far as PESQ measurements, from what I see. 
A number of people seem to be buying PESQ from the IP holders in the UK. 
It tries to make a comparative perceptual measurement, but even they 
don't claim it is an MOS measurement or a substitute for one (at least 
they didn't the last time I looked - marketing depts can change all that 
:-) ).

Regards,
Steve


Paul Vinciguerra wrote:

>Have you looked at Cisco IOS Service Assurance Agent for Voice over IP?
>
>http://www.cisco.com/en/US/tech/tk648/tk362/technologies_white_paper09186a008
>01d6b68.shtml
>
>Cisco claims it offers Voice over IP codec simulation and voice over IP 
>quality measurement (MOS and ICPIF) 
>
>I'll be testing it in the next couple of months.
>
>Paul
>
>
>On Tue, 05 Oct 2004 07:41:24 +0800, Steve Underwood wrote
>  
>
>>Hi James,
>>
>>MOS = Mean Opinion Score
>>
>>Do machines have opinions? :-)
>>
>>MOS is a costly, and definitely not an automatable, process. It 
>>requires an extensive formalised test of the system with a large 
>>number of humans. If you have seen packages which say they automate 
>>MOS testing, they are lying. Sometimes they just seem to use the 
>>term MOS for something other than a real MOS.
>>
>>I believe the R-factor, and other similar factors, are a problem for 
>>free software. Although P.853 defines a testing method, I understand 
>>there are patents on the methods used, as the wiki page mentions.
>>
>>Regards,
>>Steve
>>
>>James H. Thompson wrote:
>>
>>    
>>
>>>We are looking for automated solutions to measure call quality (MOS scores
>>>or R-Factor) to/from several points in our network.  After looking at 
>>>      
>>>
>whats
>  
>
>>>its going to cost buy an off-the-shelf system to do this, and suffering
>>>sticker shock, I thought maybe this is something that could be measured 
>>>      
>>>
>on a
>  
>
>>>call between two Asterisk systems.
>>>
>>>If anyone has suggestions, or is interesting in doing development 
>>>      
>>>
>(possibly
>  
>
>>>for $) on this let me know.
>>>
>>>Information on MOS and R-Factor
>>>http://www.voip-info.org/wiki-Call+Quality+Metrics
>>> 
>>>      
>>>