How Well Does the CI Work? - Listen to Simulations
The goal of all that applied mathematics, the signal processing, the engineering, etc., is to enable someone to hear. So how well does it work? Certainly that is a very pertinent answer, but also a little complicated. Firstly, the CI is completely dependent upon the auditory nerve. The medical team tries to make sure that the nerve has sufficient function before the implant procedure is performed. However, there is always some risk that the system will still not be able to stimulate the auditory centers of the brain. Secondly, a key component of the processing is performed by the wetware. The human brain must learn to properly interpret the incoming signals. That grey processor is quite incredible, and can be trained to an amazing level, so that it might be able to recognize sound patterns that we (as normal hearing people) might not recognize. However, the longer the duration of deafness and the age at onset of deafness affect the brain’s ability to learn these patterns. A third factor, related to the hardware, is how good the processing is and how many channels of information can be effectively passed to the cochlea. The current FDA approved devices send multiple channels of information, stimulating at several locations, in accordance with a specialized, spectrum based signal, usually optimized for speech recognition. More channels are better, but the bleed over of electrical stimulation limits the maximum number of electrodes that can reasonably be used. The state of the art processors typically employ between 12 and 22 stimulus electrodes, supporting approximately the same number of channels of information. As a general rule, the more channels, the richer the spectral content and the more recognizable the sound information.
To get a sense of the intelligibility of the signal, you should check out the audio samples below. The signals are only SIMULATIONS of what the speech processor is able to produce, without any signal degradation due to the implant interface to the auditory nerve, lack of nerve conduction to the brain, or inability of the brain to process. We thank Dr. Loizou and the University of Texas at Dallas Cochlear Implant Lab, along with Dr. Michael Dorman (Arizona State University) for providing these simulations.
# of Channels |
Processed speech |
Original speech |
1 |
|
|
For more demos and detailed information, check out http://www.utdallas.edu/~loizou/cimplants/cdemos.htm.
