Evaluation of In-Car Communication Systems - Digital Signal Processing for In-Vehicle Systems and Safety - page 99

5.5.3 Quality Degradation for the Talking Passenger
Because of the reproduced speech signal, the reverberation within the passenger
compartment is increased. This disturbs not only the listening passenger but also
the talking passenger. Therefore, the determination of the reverberation time can
be utilized to define the quality degradation for the talking passenger as well.
If the delay is too large, the acoustic wave fronts (the one of the talking
passenger and the playback of the ICC system) are perceived separately which
impairs the communication for the talking passenger. Therefore, also the delay is an
interesting quality measure.
However, if the delay is sufficiently small but the gain is too large, the talking
passengers perceive their voices as echoes. To evaluate this phenomenon, the
impulse responses between the mouth and the ears of the talking passenger are
measured. By means of these impulse responses, the corresponding frequency
responses can be calculated and evaluated. The difference between the frequency
responses of an activated ICC system and a deactivated one gives a measure of the
frequency-selective amplification of the speech signal. If this amplification is too
large, there is a high probability that the talking passengers hear themselves. In
addition, arising echoes due to the feedback can be detected by an inspection of the
impulse response. Therefore, a masking envelope can be defined and compared
with the impulse response. If the impulse response exceeds, this masking envelope
echoes should be audible [ 6 ].
5.6 Some Ideas for an Automatic Evaluation of ICC Systems
Even if we have explained a range of different objective measures which are
necessary to obtain an automatic evaluation of the ICC system, a complete defini-
tion of such an evaluation process is difficult. However, some first approaches and
some ideas for further considerations will be given in this section.
As seen in the previous sections, there are two main questions which have to be
answered when evaluating an ICC system:
1. Does the ICC system improve the speech quality for the listening passenger?
2. Is the communication quality reduced for the talking passenger by the ICC
system?
By answering these questions, a statement about the quality of the complete ICC
system can be made.
The first question can be answered by dividing this into two partial answers. The
first considers measures which indicate the improvement for the listening passen-
ger. The second partial answer deals with the factors which impair the communica-
tion quality for the listening passenger. For example, the speech intelligibility
improvement can be identified by determining the SNR improvement in the ears