Standards for measuring voice quality

March 18th, 2014, Published in Articles: EngineerIT


Hello…hello…can you hear me…hello please can you move to another spot so that the signal can improve. This type of conversation is becoming more the norm than the exception in South Africa lately. With the mobile voice penetration in South Africa being almost 100% it evident that voice traffic is aprominent medium of communication. The latest Mobility 141 research study shows that voice calls is still the most significant spend in one’s mobile expenditure.

The market prediction that networks would convert voice to data in a short time frame has not happened at the rate predicted. Voice continues and will continue to be an important means of communication long after the networks convert from circuit switched to data only transport. As bandwidth becomes more readily available and networks convert to VoIP, voice will be carried in these networks although possible in data media as VoIP and VoLTE.

Rui Godingo

Rui Godingo

As voice migrates from the traditional circuit switched based networks to data only networks the customers expect real time voice to be of identical quality to what they have been accustom to and as close as possible to the natural voice. And as a result the regulatory bodies are monitoring quality standards to measure the capability of networks to deliver premium voice. Voice quality is and will remain the major key performance indicator for the operator’s network quality. Bad voice quality represents noise, echo degraded tele-conversations, cross talk and immediately sets poor level of satisfaction

The value of voice quality is represented in MOS (mean opinion score) values and these can be interpreted on a scale of 1 to 5.

Impairment Grade
Excellent 5
Good 4
Fair 3
Poor 2
Bad 1


The existing common standards are 3SQM (single sided speech quality measure), PESQ (perceptual enhanced speech quality) and more recently POLQA (perceptual objective listening quality analysis).

3SQM is an algorithm, developed for non-intrusive voice quality testing. It is based on the ITU-T recommendation P.563. From the perspective of network operators, whose interest is in permanent network quality control, a ‘non-intrusive’ method, only based on single sided monitoring without generating traffic may sometimes be preferable.

PSQM (perceptual speech quality measure) and PESQ (perceptual evaluation of speech quality), insert a reference signal into the device under test. A stored reference is sent and the received listening quality is analysed in comparison to the original.

PESQ is an enhanced perceptual quality measurement for voice quality in telecommunications. It was specifically developed for end-to-end voice quality testing under real network conditions. It employs true voice samples as test signals and is based on the ITU-T recommendation P.862 which is the successor of ITU-T P.861 (PSQM).

POLQA (perceptual objective listening quality analysis) is the next-generation voice quality testing standard for fixed, mobile and IP-based networks that was adopted in 2011 as ITU-T Recommendation P.863. POLQA is the successor of PESQ. It avoids weaknesses of its predecessor and is extended towards handling of higher bandwidth audio signals. Further improvements target the handling of time called signals and signals with many delay variations. POLQA supports measurements in the common telephony band (300–3400 Hz), but also HD Voice in wideband and super-wideband speech signals (50–14000 Hz)

POLQA covers a model to predict speech quality by means of digital speech signal analysis. The predictions of those objective measures should come as close as possible to subjective quality scores as obtained in subjective listening tests. POLQA, like PESQ, uses real speech as a test stimulus for assessing telephony networks (2).

The common idea behind perceptual quality measures is to mimic the situation of a subjective test, where human beings would have to score the quality of sound samples in a listening laboratory environment. The result is called mean opinion score (MOS). For a long time, subjective test procedures were the only means of assessing the sound quality impression. Of course, subjective experiments require a huge number of subjects to achieve statistically relevant results, and thus are very costly and time consuming. For quality testing in the laboratory as well as in the field, subjective tests are clearly no option. Consequently, perceptual models can be applied to generate an objectively derived quality metrics (objective MOS or OMOS) that can be compared to a mean opinion score scale. In the course of perceptual processing, many other detailed models output values, such as spectra, dynamically measured bandwidth, distortions, modulation etc. are generated and reported to make this technology universally applicable.

Maintaining the standards of service quality remain challenging. Reaching a seamless, 24/7/365, standardised quality supervision across all technologies and services can be time-consuming. Automated test methods with the intention to mimic subjective tests, allow for a cost effective and consistent alternative to subjective testing, either on a referenced or unreferenced algorithm.

Saab Grintek Technologies in partnership with Keynote SIGOS offers a wide range of test cases to verify quality of service for network, roaming and international carrier for radio, legacy fixed and IP testing. The SIGOS integrated test environment (SITE), is the most advanced active end-to-end test system available on the market. It enables network operators and mobile service operators all around the world to host and work with their own testing facilities whilst easily integrating into existing network infrastructures. Additionally, the partnership has the capability to offer a worldwide leading reference system for automated outbound roaming testing called “GlobalRoamer” (3).

The South African ICT regulator ICASA has been reporting more frequently on the call quality by publishing statistics for drop call rate (DCR) and call set-up success rate (CSSR) which implies consumers are becoming more demanding and expect a high quality service in return for the funds spend.



[2] Opticom – Typical PESQ results for various ITU-T and GSM standardised codecs, OPTICOM GmbH, based on ITU-T Recommendation P.862.3 Appendix I, 2007-02-23, v1.0


Contact Rui Godinho, Saab Grintek, Tel 012 672-8225,>

Related Articles

  • Now Media acquires EngineerIT and Energize from EE Publishers
  • Printed electronics: The defining trends in 2019
  • Charlie and the (fully-automated) Chocolate Factory
  • SANSA app calculates best HF communication channel
  • New app supports predictive services for drive systems