Next: The specific bimodality
Up: The bimodality of
Previous: The bimodality of
Setting communications
parameters at threshold level, Risberg and Lubker observed that when a
speaker appeared on a video display, but with the sound turned off,
subjects relying on speech-reading correctly perceived 1% of test
words [290]. When the subjects could not see the display, but
were presented with a low-pass filtered version of the speech sound, they
got 6% correct. Presented with the combined information channels, the
performance jumped to 45% correctly perceived test words. This
observation exemplifies the remarkable synergy of the two modes of speech
perception. However, little is known about the process that integrates the
cues across modalities, although a variety of approaches and of models to
multimodal integration of speech perception have been
proposed [208,332,40] and
tested [211,213,128,210,291]. Our understanding of
this process is still relatively crude, but its study is very active and
controversial at present (see the 21 remarks to and in [209]!). For
an extensive presentation of the various integration models proposed and
tested in literature, see the ICP-MIAMI 94-2 Report by Robert-Ribes.
Esprit Project 8579/MIAMI (Schomaker et al., '95)
Thu May 18 16:00:17 MET DST 1995