Evaluation of In-Car Communication Systems - Digital Signal Processing for In-Vehicle Systems and Safety - page 73

Seat-specific
processing
(for talking
passengers)

Seat-specific
processing
(for listening
passengers)

Preprocessing
and analysis
filter banks

Loudspeaker-
specific
processing

Postprocessing
and synthesis
filter banks

Loud-
speaker
signals

Micro-
phone
signals

Uncorrelated
parts of
loudspeaker
signals

Preprocessing
and analysis
filter banks

Processing
that is applied
to all microphones
that record talking
passengers

Fig. 5.5 Basic signal processing units of in-car communication systems

(e.g. noise suppression) and to all output spectra (e.g. equalization). Thus, we have
inserted two appropriate signal processing blocks in Fig. 5.5 . Some further infor-
mation about these two signal processing parts is presented in Sects. 5.3.2 and 5.3.5 .
All signal components involved in between the microphone and the loudspeaker
signal enhancement units can be grouped into two conversion parts. One group has
to extract a dedicated signal for each seat that is occupied by a talking passenger.
This can be done simply by selecting one of the microphones or by combining
several of them (e.g. by means of beamforming). The outputs of this first signal
processing group will be called the seat-specific signals and parameters of the
talking passengers. The second group takes these talker signals and maps them
onto signals that are specifically designed for the individual listeners. This includes
a mixing process as well as gain adjustments in dependence of the noise level
estimated at each listener seat. Finally, each listener signal is mapped onto the
loudspeakers that are assigned to the listener seat using appropriate gain and delay
settings.
In comparison to hands-free telephones or speech recognition engines, no
methods for evaluating the quality of ICC systems have been standardized yet,
and only a few have been published (e.g. [ 6 , 7 ]). Thus, evaluation is more challeng-
ing as in most other speech and audio applications. Considering a basic ICC system,
we focus on the analysis of ICC systems and give a general idea of how automatic
evaluations can be performed.
Before we start with the evaluation, the restrictions and the boundary conditions of
ICC systems as well as their consequences are discussed in the next section. Since a
basic understanding of the involved signal processing units of a system is necessary
in order to be able to design appropriate measurements and tests, we will continue with
brief summaries of the individual processing units shown in Fig. 5.5 in Sect. 5.3 .
When evaluating ICC systems, usually comparisons between the communication
with and without the support of ICC systems are investigated. In addition, evaluation