Introduction

Basics of Sound

Amplitude Modulation

Frequency Resolution

Stimulation Type

References

 

 

   

 

---------------------------------------------------------------------------------------------------------------------------

INTRODUCTION TO SPEECH PROCESSORS

The speech processor transforms the electrical signal received from the microphone into stimuli for an implanted electrode array. Until recently, size limitations required that the speech processor be contained within an enclosure worn on the belt or placed in a pocket. In current cochlear implant systems, the processor may be located in the "behind the ear" (BTE) housing, as shown in the illustration above. Despite the smaller size and convenience associated with a BTE housing, speech processors enclosures, which range roughly from the size of a walkman to the size of an iPod Shuffle, can execute more complicated speech processing algorithms than processors in the BTE housing. Nevertheless, speech processors contained in the BTE housing have similar capabilities to body-worn processors and so are rapidly supplanting body-worn processors. (Wilson, 2004)

In systems containing a separate speech processor enclosure, the processor's input and output are sent via different wires in a single cable that connects the speech processor to the BTE housing. A second cable transfers the output of the speech processor from the BTE housing to the transmission link. In systems where the speech processor is contained within the BTE housing, only one cable is involved in sending the output of the processor to the transmission link. (Wilson, 2004)

The speech processor is powered by hearing aid batteries for head-worn processors in the BTE housing and AA batteries for body-worn processors. The typical battery life exceeds 12 to 16 hours and allows devices to be used during the daytime without recharging or replacement. This adequate battery life is largely a result of the use of low-power digital signal processing (DSP) chips. Cochlear implants benefit from the rapid advances in mobile technologies, such as cell phones, that utilize DSP chips. (Wilson, 2004)

 

Back to top

 
             

---------------------------------------------------------------------------------------------------------------------------

Mechanics of Sound

Sounds are pressure fluctuations that result from the movements of air molecules due vibrations in the air and can be represented as a wave form that traverses through time and space (Luther, 1999)

         
Figure 1 - A simple waveform of sound as it traverses through time. T indicates the wavelength. (Obtained from Luther, 1999

The figure above shows a simple illustration of a sound wave as it moves over time. However, a real world auditory environment, wave forms are not as simple. Instead, most auditory waveforms are actually very complex auditory waveforms (shown in black) that result from the summation of many different waveforms (red, blue, and green waves) (Wright, 2000).

     
  Figure 2 - An example of the summation of many waveforms to create a complex auditory waveform. The waveform in the top panel is actually the resulting summation of the waveforms on the bottom panel. (Obtained from Wright, 2000).
         
                       

It is the job of the signal processor to decompose this complex auditory signal received from the microphone into discrete electrical pulses to be sent to the auditory neurons via the transmission link (transmitter and receiver) and electrodes. This is done via a variety of methods, including amplitude modulation, frequency decompositions, and spectral analysis (Rubenstein, 2004; Wilson, 2004; British Cochlear Implant Group, 2006).

Back to top

     
---------------------------------------------------------------------------------------------------------------------------

Amplitude Modulation

To obtain signals within the appropriate dynamic range of hearing, signals must be compressed and limited (British Cochlear Implant Group, 2006). Figure 3 illustrates this strategy, both individually and in combination. As shown in the top left graph of this figure, the signal is limited by not processing the lower and higher levels of sound (British Cochlear Implant Group, 2006). The top right graph shows that by compressing the signal, the higher levels of sound are attenuated (British Cochlear Implant Group, 2006). The bottom graph shows the combination of compression and limiting, which results in a signal being in the desired dynamic range (British Cochlear Implant Group, 2006). All speech processors in cochlear implants employ these types of strategies to alter the auditory waveform heard.

             
      Figure 3 - Illustration of limiting and compression strategies to a signal. (Obtained from British Cochlear Implant Group, 2006).
           
 

Back to top

 

---------------------------------------------------------------------------------------------------------------------------

Frequency Resolution

Prior to amplitude modulation, the speech processor splits the sound waveform "heard", into its frequency components. This is done one of two stimulation schemes as illustrated in Figure 4, analog or pulsatile processing; analog processing sends a signal that is continuously varying in time and amplitude to the implant, whereas pulsatile processing sends a train of pulses (usually biphasic) (Rubenstein, 2004; Wilson, 2004, Koch, 2000). Analog stimulations stimulate electrodes in a simultaneous fashion, where all necessary electrodes are stimulated at one time, whereas pulsatile stimulations stimulate in a non-simultaneous or sequential fashion, where only one electrode is stimulated at any one time (Rubenstein, 2004; Wilson, 2004, Koch, 2000).

Figure 4 shows how regardless of the processing strategy, the sound waveform is separated by band-pass filters into different frequency components; each band-pass filter passes only frequencies within a certain range and rejects all other frequencies (Rubenstein, 2004; Wilson, 2004; Wikipedia contributors, 2006). The resulting signals from the band-pass filtering are each compressed so that the amplitude falls within the desired dynamic range (Rubenstein, 2004; Wilson, 2004).

   
  Figure 4 - A) Typical scheme of analog processing where compressed output signals of the band-pass filters are sent directly to the electrodes for simulation of the auditory nerve. B) Typical scheme of pulsatile processing where outputs of band-pass filters have envelopes extracted and then compressed. Signals are then converted to pulses which are sent to the cochlear in either an interleaved fashion. (Obtained from Rubinstein, 2004)

Back to top

---------------------------------------------------------------------------------------------------------------------------

Stimulation Types

The receiver detects the amplitude variations in the sent signals at specific frequencies and generates electrical signals that are sent to various electrodes on the array simultaneously (British Cochlear Implant Group, 2006). The drawback for this method is that because electrodes are all stimulated at the same time, there is a potential for one stimulated electrode to interfere with the stimulation of another electrode (British Cochlear Implant Group, 2006).

Unlike analog processing, pulsatile processing employs trains of biphasic pulses that are sent to the electrodes in a non-overlapped form, which minimizes electric field interaction between stimulated electrodes, which can lead to disruption (British Cochlear Implant Group, 2006). The signal is passed through a series of band-pass channels, where each channel is made up of a band-pass filter, envelope detector, and modulator (British Cochlear Implant Group, 2006; Koch, 2000). The envelope detector generates the envelope of the decomposed signal using a rectifier and low-pass filtering (Rubenstein, 2004; Wilson, 2004). The rectifier converts the AC signal into a DC signal, from which the peak level/amplitude of the input signal can be obtained (Wikipedia contributors, 2006). The rectifier generates new frequency components, some of which are unwanted by that channel; thus, the low-pass filtering removes these unwanted components from the resulting signal (Rubenstein, 2004; Wilson, 2004; Wikipedia contributors, 2006).

The envelope detection step in pulsatile processing allows for the use of the entire spectrum of incoming sound, and thus, the temporal information in the sound being heard is not compromised (Rubenstein, 2004; Wilson, 2004). As Figure 5 shows, a sound wave (frame on left) can be decomposed into an envelope (top frame on right) and a temporal structure (bottom frame on right) (Rubenstein, 2004; Wilson , 2004). The temporal structure relays important information regarding the pitch, timbre, and timing of the sound; different pulsatile strategies employ this principle in different ways to process sound (Rubenstein, 2004; Wilson, 2004).

     
    Figure 5 - The frame to the left shows the acoustical signal for the word "boy." The top frame on the right shows the envelope of this signal, while the bottom frame shows the temporal structure of this signal. (obtained from Rubinstein, 2004)
   
     

Back to top

---------------------------------------------------------------------------------------------------------------------------

References

British Cochlear Implant Group (2006). Speech Processing Strategies. British Cochlear Implant Group. Retreived April 24, 2006 from http://www.bcig.org/professional/allprofs_strat.htm .

Clark G (2004). Cochlear Implants. In Greenberg S, Ainsworth WA , Popper AN, Fay RR, eds. Speech Processing in the Auditory System. New York : Springer, 422-462.

Koch DB. (2000). Cochlear Implants: An Overview. Healthy Hearing. Retrieved April 20, 2006 from http://www.healthyhearing.com/library/article_content.asp?article_id=43 .

Luther BA (1999). Wave Motion: Physics 128 Lecture. Concordia College . Retrieved April 26, 2006 from http://www.cord.edu/dept/physics/p128/lecture99_33.html .

Rubenstein JT. (2004). How Cochlear Implants Encode Speech. Current Opinions in Otolaryngogology and Head and Neck Surgery . 12, 444-448 .

Wikipedia contributors (2006). Band-pass filter. Wikipedia, The Free Encyclopedia. Retrieved April 24, 2006 from http://en.wikipedia.org/w/index.php?title=Band-pass_filter&oldid=49725528. Wikipedia contributors (2006). Rectifier. Wikipedia, The Free Encyclopedia. Retrieved April 24, 2006 from http://en.wikipedia.org/w/index.php?title=Rectifier&oldid=49796549 .

Wilson BS. (2004). Engineering Design of Cochlear Implants. In Zeng FG, Popper AN, Fay RR, eds. Cochlear Implants: Auditory Prostheses and Electric Hearing. New York : Springer, pp. 14-52.

Wright EL(2000). Anomalous Dispersion, not Faster than Light. UCLA Astronomy. Retrieved April 26, 2006 from http://www.astro.ucla.edu/~wright/wave-packet.gif .

Back to top