UMTS Speech Service

Although UMTS will be used for a variety of data services, speech may well
remain the most widely used service. Speech has certain requirements in
terms of data rate, delay, jitter, and error-free delivery, all of which are
derived from human perceptions and expectations. Moreover, speech quality
in UMTS needs to be comparable to that in fixed telephony networks
and certainly no worse than that experienced in 2G wireless networks.
UMTS uses the Adaptive Multirate (AMR) speech coder. This is actually
several coders in one and provides coding rates of 12.2 Kbps, 10.2 Kbps, 7.95
Kbps, 7.40 Kbps, 6.70 Kbps, 5.90 Kbps, 5.15 Kbps, and 4.75 Kbps. The 12.2-
Kbps rate is the same coding scheme as used in the GSM Enhanced Full-
Rate coding scheme. The 7.4-Kbps rate is the same coding scheme as used
in IS-136 TDMA networks. The reuse of existing coders means that the
voice-coding scheme of UMTS should at least offer the same levels of quality
as experienced in existing 2G networks.
The AMR coder allows for the speech bit rate to change dynamically during
a call. As we shall describe later, the higher the bit rate of any service,
the smaller the effective footprint of a cell. Thus, a user at the edge of a cell
could change from a high speech-coding rate to a lower speech-coding rate
to effectively extend the coverage for speech service. Each AMR speech
frame is 20 ms in duration and it is possible to change the speech-coding
rate from one speech frame to the next. Thus, the coding rate could change
as often as every 20 ms, although that is unlikely to ever happen in reality.
The AMR coder also supports voice activity detection (VAD) and discontinuous
transmission (DTX), with comfort noise generation. The net effect is
that little or nothing is sent over the air interface when nothing is being
said. Given that typical speech involves one person speaking, followed by
the other, it is possible to reduce the amount of transmission over the air
interface by as much as 50 percent. Of course,VAD and DTX are supported
by most modern wireless technologies.
Many of the services supported by UMTS are packet-switched data services.
Speech, on the other hand, at least in 3GPP Release 1999 and 3GPP
Release 4, is a circuit-switched service. This means that a user in a speech
call has access to dedicated resources throughout the call. In effect, a dedicated
pipe is used between the two parties in a speech conversation. This is
similar to the way speech is handled in a GSM/GPRS network, where a
speech call uses a dedicated timeslot on the air interface and uses a dedicated
transport and switching in the core network. Although the concept of
timeslots does not map well to WCDMA radio access, the assignment of
dedicated resources still applies.