Understanding the Requirements for Toll-Quality Voice Chapter 9: Network Requirements and Preparation
9.4.11 Echo Cancellation
Echo in a voice communication system is caused by signal reflections generated by the
electrical circuits called hybrids that convert between two-wire (shared transmit and
receive pair) and four-wire circuits (separate transmit and receive pairs). These reflections
cause the speaker’s voice to be heard in the speaker’s ear as delayed by many milliseconds.
Echo is present even in the traditional circuit-switched telephone network, but since the
delay in a local circuit-switched call is so low, the echo is not perceivable. On a packet-
based voice network, there is more delay, and the speaker may perceive the echo if it is not
The DSP software on the ShoreGear voice switches provides dynamic echo cancellation.
When a user places an extension-to-trunk call using an analog trunk on a ShoreGear voice
switch, the user’s voice bounces off the initial four-wire to two-wire conversion in the
analog trunk circuit, then off the two-wire to four-wire in the central office, and finally off
the called party’s telephone. This echo returns from the central office and is cancelled by
the echo canceller on the trunk port of the voice switch. The echo from the called party’s
phone, however, is usually cancelled or suppressed by the central office. If this echo is not
cancelled, the user may hear himself or herself talking.
In the opposite direction, the external person’s voice bounces off the user’s telephone. This
echo returns from the telephone and is cancelled by the echo canceller on the telephone
port of the voice switch. If this echo is not cancelled, the external party hears himself or
herself talking. This same process of echo cancellation applies to extension-to-extension as
well as trunk-to-trunk calls.
ShoreGear switches can cancel echo received up to 16 msecs after being sent.
9.4.12 Resultant Voice Quality
As stated earlier, the ShoreTel 8 system has been recognized for excellent voice quality. This
is a result of the excellent hardware and software design that minimizes latency and
dynamically adapts to the effects of jitter, packet loss, and echo introduced by the network.
There are two subjective testing methods that are used to evaluate voice quality. A method
called Mean Opinion Score (MOS) is an open test in which a variety of listeners judge the
quality of a voice sample on a scale of 1 (low) to 5 (high). There is general industry
agreement on the theoretical maximum MOS value on a per codec basis that can be
achieved (see Table 9-9).
Both the MOS test method and an interactive test method were used by Miercom. The
interactive test focused on the conversational quality of the call. The results are shown in
Table 9-10. The ShoreTel MOS scores are higher than the industry-standard values. This is
likely a result of the subjective nature of the head-to-head test, which scores a relative
ranking rather than an absolute ranking.
Codec Data Rate (Kbps) MOS
Linear 128 4.5
G. 7 11 6 4 4 . 1
ADPCM 32 3.85
G.729a 8 3.85
Table 9-9 Theoretical MOS Maximum Scores