User guide

Aries Reference Manual Section 2 – Viewing and Analysing Data
Radio Systems Information Ltd. Page 41
3 Voice Quality Testing
Digital radio networks use speech vocoders and require sophisticated measurement techniques using
human speech samples to provide a true and repeatable Mean Opinion Score (MOS). Aries achieves this
by using a digital signal processing algorithm with an auditory model that takes into account the
psychophysical properties of human hearing and the subjectivity of the distortion in the received signal.
This means that the measured MOS will truly reflect the speech quality perceived by the user and will
typically be within a half point of a MOS score determined by a controlled subjective test in a laboratory.
The algorithm used in Aries is the industry standard known as 'PESQ' (Perceptual Evaluation of Speech
Quality) meeting ITU recommendation P.862.
Conventionally Bit Error Rate (BER) and signal strength have been used as the prime indicators of radio
system quality however, now that objective methods for measuring speech quality are available, it makes
more sense to use speech quality as the prime indicator. BER and signal strength have the following
limitations:
• Voice quality is the only parameter to compare networks of different technologies.
• Voice quality is measured in both the uplink and downlink whereas BER and signal strength are
normally only measured on the downlink.
• Signal strength used on its own is not always a good indicator of performance for digital networks.
Multi-path effects that cause inter-symbol interference may mean that a strong signal could have a
poor BER.
• BER does not remain at a constant level during a call hence laboratory measurements of vocoder
speech quality at a constant BER are not always representative of the speech quality produced by
a signal with a fast changing BER.
• BER does not detect failures within voice processing circuits.
• BER does not detect echo or other audio effects caused by the PSTN.
Voice Quality testing using the PESQ algorithm has therefore become the de facto standard methodology
for testing TETRA network performance.