Question 9/12 - Objective measurement of speech quality under conditions of non-linear and time-variant processing

Français Español

Print Version

Home : ITU-T Home : Study Period 2001-2004

(Continuation of parts of Question 13/12 studied in 1997-2000)

Type of question

Task-oriented, leading to new and revised ITU-T Recommendations.

Background

During the previous Study Periods, several methods for objectively evaluating non-linear and time-variant processes have been studied. This led to an agreement on a single method, based on the Perceptual Speech Quality Measure (PSQM), which is contained in Recommendation P.861. Recommendation P.861 specifies a measure for objectively testing voice-band (300 - 3400 Hz) coders under the varying conditions of transcodings, codec speech input levels, talkers, languages, and bit rates.

Since Recommendation P.861 was recognized to have certain limitations in specific areas of application, it was developed an enhanced version of P.861 by the end of the 1997-2000 Study Period, covering a wider variety of conditions under which telephones are used and where more robust and more reliable methods for objectively assessing the speech quality are needed. The development of this new Recommendation was completed in 2000. It was determined as ITU-T Recommendations P.862 and will replace P.861 early in 2001.

Although it could be envisaged that P.862 covers a much wider range of speech transmission conditions than P.861, it is still limited to assess the speech quality of end-to-end narrow-band handset telephony. In the near future, there is envisaged a strong need for objective quality assessment methods of speech signals in complex transmission scenarios, particularly in telecommunication systems with modern transmission technologies and new types of terminals.

Text of the question

Considering the need for objective measures that work in telecommunication systems with modern transmission technologies and new types of terminals under a wide variety of conditions:

1) Can P.862 be enhanced to operate under a wider variety of conditions? Examples include, but are not limited to:

wideband signals
talker dependencies
effects of non-linear processing in echo cancellers

2) How can P.862 be extended to mouth-to-ear transmission scenarios including modern terminals for mobile and/or hands-free applications?

3) How can psycho-acoustically based speech quality assessment methods like P.862 be combined with analytical measurement methods for specific signal parameters?

4) How can psycho-acoustically based speech quality assessment methods like P.862 be improved by using additional information derived from network signalling or transmission protocols?

5) How can users of speech quality assessment methods like P.862 be guided for correct application and interpretation of the obtained results?

6) How can psycho-acoustically based speech quality assessment methods be extended for single-ended objective speech quality assessment algorithms.

Study items

The main points of study are:

To extend objective speech quality assessment algorithms to wideband telephony;
To develop objective speech quality assessment algorithms for ‘talker quality’;
To develop single-ended objective speech quality assessment algorithms;
To improve P.862 according to evolving speech transmission technologies;
To apply objective speech quality assessment algorithms based on psycho-acoustic principles like Recommendation P.862 to mouth-to-ear transmission scenarios, i.e. including modern terminals for mobile and/or hands-free applications;
To define methods for assessment of parameters like ‘conversational quality’, ‘double-talk performance’ with non-linear and time-varying behaviour of network components or terminals;
To combine information of both, the speech signal as well as the data streams (in digital transmission systems) for adaptive quality/performance control of services, networks and/or terminals;
To provide guidance to inexperienced users of objective speech quality assessment algorithms in order to prevent inappropriate use or misinterpretation of results.

Objectives and schedule

It is anticipated that new and/or updated recommendations will be developed by the end of the study period.

Relationship with other Study Groups

Most of the activities relating to the Question are likely to be internal to ITU-T Study Group 12 (e.g., SQEG), although close liaison with ITU-T Study Groups 2, 9, 15, 16, SSG and relevant ITU-R Study Groups is desirable to ensure that the Recommendations produced under this Question can be used effectively to study the performance of networks and terminals, respectively, generated by these bodies. In addition, coordination with other standardization activities, e.g. in ETSI TC STQ or ETSI EP TIPHON will ensure that Recommendations developed in this Question satisfy the user’s needs.