Question 7/XII - Models for predicting transmission quality from
objective measurements
(continuation of Question 7/XII studied in 1985-88)

Considering

(a) that according to CCITT Recommendation P.74, "Conversation tests"
are preferable to other methods for subjective determination of
transmission quality;

(b) that such tests cannot be carried out on a sufficiently large scale
to cover the whole range of all degrading factors that are of interest
and combinations of more than two factors at a time are very rarely
feasible;

(c) that Supplement No. 3, to CCITT Recommendation P.11 contains
descriptions of models used respectively by Bellcore, BT, CNET and NTT
for assessing the effects of customer satisfaction of certain ranges of
degrading factors;

(d) that, in principle, it is desirable that the following degrading
factors should be taken properly into account by such models:

Factor 1. Transmission loss

2. Circuit noise

3. Room noise

4. Attenuation/frequency distortion

5. Sidetone

6. Non-linear distortions

7. Echo

8. Propagation time

9. Delay distortion

10. Mutilation by voice switching;

CCITT decides to study the following Question (the Question consists of
three parts):

Part 1 - What changes can be made to the models to incorporate
additional degradations?

Part 2 - How well do all these models predict the results of
"Conversation tests" conducted with real subjects?

Part 3 - To what extent are the predictions from the different models
compatible with each other?

Note 1 - In studying the Question, comparisons have to be made:

i) between the results of "Conversation tests" as described in
Recommendation P.74 conducted by different administrations;

ii) between predictions by different models, and

iii) between a particular set of "Conversation test" results and a
particular model.

These comparisons have been hampered because uniform methods have not
been used for expressing the objective characteristics of the
connections. It is strongly recommended that as much use as possible be
made of methods defined in CCITT Recommendations and relating to
loudness ratings, especially P.64 and P.76.

Note 2 - A great deal of information helpful in studying this Question
appears in contributions to other Questions, especially 9/XII
(Sidetone), 13/XII (Non-linear distortion) and 18/XII (Quantizing
distortion).

Note 3 - Annex 1 contains the text of the reply to the Question after
study during 1985-1988. Annex 2 contains a list of contributions which
have become available for use in the study of the Question.

Note 4 - It is intended that the model or models finally recommended
shall be used to update the numerical values now given provisionally in
Recommendation P.11. The model or models may also be used for studying
the Questions mentioned in Note 2.
ANNEX 1

(to Question 7/XII)

Refer to the list of documentation in Report COM XII-R 29,
February 1988, sections 4.1 and 4.2
ANNEX 2

Refer to the reply to Question 7/XII in Report COM XII-R 29,
February 1988, section 4.4
ANNEX 3

(to Question 7/XII)

It seemed to be urgent to prepare a Recommendation for the quickly
developing land mobile radio-telephone service. The aim was to have
transmission quality equal or at least very similar to the PSTN. On the
other hand due to the limited [25 kHz, 30 kHz] radio frequency bandwidth
16 kbit/s voiceband coding is necessary. To meet simultaneously the
above mentioned requirement was up to now not possible. Therefore
further study is necessary which must be supported by