In an exemplary embodiment, the point of departure is that of a substitute signal system for deaf individuals which codes acoustical signals, in particular speech sound, into electrocutaneous stimulation patterns. Current pulse patterns serve as substitute signals which, via skin surface electrodes, e.g. on the forearm, bring about sensory patterns which can be interpreted as speech information. To this end, the sound signal is separated with band pass filters into e.g. twelve component signals characteristic for the recognition of speech, whose levels are converted into pulse durations, whereby the dependency of the sensation intensity upon the stimulation parameters is taken into account. The mean frequencies of the band pass filters are associated with a corresponding number of locations on the forearm (i.e. in the present example twelve). A simulation of lateral inhibition suppresses irrelevant spectral components and thus acts in a selectivity-increasing (or enhancing) fashion. It is thereby possible that, after only approximately fifteen hours of practice, ten monosyllabic words are correctly recognized with a reliability of approximately 95% by means of the sensor patterns which are to be transmitted in accordance with the disclosure. The disclosed method and apparatus for its realization are particularly suited for use in hearing prostheses for deaf individuals.
Read full abstract