A speech treatment algorithm based on a programmable filter bank for cochlear prostheses

In this article, a programmable stimulation algorithm developed for cochlear prostheses has been described. The algorithm was based on a speech treatment system with programmable filter-bank modeling which allowed flexible speech signal processing. With its digital system, it could be adapted to any cochlear prosthetic apparatus driven by a Digital Signal Processor (DSP). The different bands of the digital filters used in the speech-processing of this stimulation algorithm were fully programmable, and could be adjusted to the patient's pathology. This provided clinicians with the possibility of dividing up the frequency band to adapt the filtering according to the patient's audiogram and different pre- and post-operative tests. To generate stimuli in the nerve endings of the cochlea, the proposed algorithm allows the speech signal energy to be extracted by the filter bank (f1, f2 ..., fN), coding of the output information to be made, and the appropriate commands to be transmitted to the internal part of the cochlea via the prosthetic modulator. In fact, the speech-signal energy (E1, E2,..., EN) extracted during the processing phase serves to estimate the stimulating pulse parameters transmitted to the inner ear (cochlea) through the implanted micro-stimulator. An IBM compatible personal computer (PC) is connected to the external part of the prosthesis to allow clinical assessments and adjustments to be made of the different parameters, especially during the first clinical experiments. On the other hand, this algorithm was developed to generate a real-time display showing the affected frequency bands as well as their relative stimulation levels. This can assist patients during the rehabilitation phase by providing a visual reference. Hence, with this visual tool, clinicians are able to carry out therapeutic experiments during the rehabilitation phase, correctly adjust the operational parameters of the hearing device, and assess the electrical charges to be injected into the cochlear biological tissue (current pulse levels). The software provided in this stimulation algorithm ensures flexibility in programming, ease of use, and different safety features that can help to satisfactorily meet individual needs. (C) 2000 Editions scientifiques et medicales Elsevier SAS.