There are some attempts at standardizing the measurement of voice quality. One of the best known is the MOS scale that is based on subjective measurements. We can also relate the E model since it brings in some objective parameters like network delay and packet loss. Something interesting about this last model is that it contemplates the conversion of MOS scale results, which allows us to obtain a standard scale to quantify voice quality.
The MOS scale is really a recommendation of the ITU. Specifically the ITU-T P.800 recommendation. This scale writes down a voice quality scale based on the subjective samples that are realized through a series of techniques known as Absolute Category Rating (ACR).
For this, a group of people is brought together and they are asked to rate voice quality in a subjective way. Before starting the evaluation, they must listen to some previously defined examples from the recommendation so that the users have a reference frame.
Once this is done a series of phrases is transmitted (they are also pre-defined by the recommendation) through the telephone line and the users proceed to rate the voice quality.
The following is a summary of the MOS scale.
MOS rating Quality Effort:
5 Excellent No effort needed
4 Good It is necessary to pay attention but no significant effort is needed.
3 Acceptable Moderated effort
2 Poor Great effort
1 Bad Can’t be understood
A disadvantage of using the MOS scale is without a doubt the quantity of time necessary to determine voice quality in a simple line. Imagine trying to coordinate a series of tests with a great amount of people whom we would have to previously train just to evaluate voice quality of a single line.