[Asterisk-Users] How to learn or teach VoIP QoE

Steve Underwood steveu at coppice.org
Tue Sep 19 10:41:31 MST 2006


Mojo with Horan & Company, LLC wrote:

> I've heard a MOS of 4.4 is ulaw/alaw, presumably like a local exchange 
> call through the pstn.
>
> testyourvoip.com tells me that the highest score available with G.729 
> is 4.2, which is pretty darn close to 4.4.

Alaw and ulaw are about 4.4. ulaw on a robbed bit trunk is about 4.3. 
G.729 is about 3.9. That doesn't sound like a big spread, but MOS tends 
to bunch them together. G.729 at 3.9 is pretty reasonable, while 
anything scoring 3.0 is bloody awful.

>
> I don't know why I think this (or why I've heard it (or if it's 
> right)) but I think gsm is 3.8?  Maybe someone can confirm or 
> disconfirm this. This kinda seems like the codec my long distance 
> calls go out on but I'm not really sure.

3.8 sounds about right.

>
> http://www.testyourvoip.com/results.html?id=071GM0&result=0
> (one of my more POOR results) approximates about 2.75 for 'tin cans 
> and string', 3.2 for a crummy cell phone call, about 3.9 for a decent 
> cell call, 4.4 or so for 'like calling next door', and the mystical 
> 5.0 for 'better than being there'.
>
> I wonder if 16kHz wideband codecs would bring our voice-carrying 
> experiences into the 5.0 range?

Much of the reason ulaw/alaw is way below 5.0 is due to bandwidth 
limitations. 16kHz sampling gets much closer to 5.0 It takes more like 
32kHz sampling to actually reach 5.0, though.

Steve




More information about the asterisk-users mailing list