RESNA 26th International Annual Confence
The purpose of this paper is to initiate technical discussions concerning the potential uses of voice activation and how it could be used to empower individuals with physical and visual impairments. Specifically it deals with one approach of using voice recognition to control a tape deck. The device was designed for the Talking Library, which is run by Perkins School for the Blind located in Watertown Massachusetts.
The voice activated tape deck controller was designed to provide hands free operation of the Talking Libraries tape decks which are loaned out, along with books on tapes, through a mail order system to patrons who meet the eligibility requirements. Primarily the patrons of the Talking Library consist of individual with visual and or physical disabilities.
The Talking Library currently has some methods of assisting patrons with the operation of the tape deck but they are limited. A remote control can be attached to the tape deck but it's only function is to turn the power on or off. Thus the tape would already have to be in the play mode for it to start playing when the remote turned it on. The Talking Library also has a mechanical adapter that makes use of levers of different lengths to ease in the operation of the tape deck. Both of these methods although helpful for some of the patrons do not adequately meet the needs of the visually impaired or patrons with limited or no motor skills such as paraplegics, or quadriplegics. For many of the patrons the act of enjoying a good book is either impossible or so difficult that the joy usually obtained is lost by the difficulty it presents. A voice activated controller that could operate all of the functions of the tape deck would give the user complete control over the tape deck and would eliminate the need for outside assistance after the initial setup. The voice activated tape deck controller was designed to eliminate these difficulties. I
Figure 1.1. shown below depicts the voice activated tape deck controller positioned to control the tape deck that is supplied by the Talking Library. The design criteria for the controller was driven by the both the devices it would be interacting with, and the individuals that would be operating it. Based on the criteria the controller required four major sections.
The first section consists of the Voice Direct 364, a commercially available voice recognition chip produced by Sensory Inc. The voice chip has the ability to store a limited number of voice commands into its memory. The voice chip creates templates by sampling audio sounds during its training mode. Once trained the chip can then be set into the listening mode where it will sample sound from it's environment and compare it to the stored templates. When a match is found, an eight-bit output, corresponding to the stored template, will be transmitted out of the voice chip.
The second major section of the controller is the control circuitry. The voice recognition chip has an eight bit output (but only 16 distinct outputs) and the tape deck has five function keys, stop, rewind, play, fast forward, and eject. Thus it was necessary to use digital logic circuitry to convert the chips outputs to ones that could be used to control the five function keys. The digital logic is also necessary to overcome two unfavorable behaviors of the voice chip. During initial power up and any time the reset switch is activated the voice chip will set all of its outputs high for one second. This would result in all function keys being activated at the same time, which could in turn damage, the tape player. Figure 1.2 shown below depicts the logic circuitry that is used to drive the stop function of the tape player. This same design is used to control the other four functions with the only difference being the order of the inputs.
The output of each of the five sections of control circuitry is connected to a DC solid state relay which when biased by the control circuitry will switch power to the third major section of the controller, a bank of solenoids. Push type solenoids are positioned over the keys of the cassette player. When the solid state relay is biased 12 volts @ 8 amps is supplied to the solenoid providing the 96 watts required to product the force necessary to activate the key.
The final requirement was the power source. The power source consists of two components. The first component is a dual output switching power supply. The power supply has a 5 volt @ 3.5 amp output, which powers the TTL logic and the voice chip. The second output is rated at 12 volts @ 4 amps. Due to size and cost constraints it was not practical to purchase a power supply that could supply the entire 12 volts @ 8 amps to the solenoids. To work around this problem a combination of a reduced power supply in parallel with an 8200 microfarad capacitor was used. The capacitor will be charged by the power supply and when discharged will produce the required energy to drive the solenoid. Using this method I was able to reduce the physical dimensions of the power supply from 10x7x3 inches to 5x3x1.5 inches. The power supply being the largest single component in my device drove the minimum size requirements of the entire unit, thus reducing the size of the power supply allowed me to reduce the overall size of the controller. Using this method also resulted in a substantial cost reduction of $110.00 or approx. of the total $410.00 required for all parts and materials.
The controller, after initial positioning, provides the user with the ability to control the tape deck without any physical intervention. Voice commands can easily be programmed in minutes making training a simple process. The only negative issue that I observed is that increasing the volume of the tape player interferes with the ability to stop the tape player once in the play mode due to the random nature of the background noise while the voice chip is trying to recognize commands. A potential solution to this could be the use of headphones but this would require additional physical intervention.
Although the controller as designed met the requirements of my initial proposal, I have several recommendations for any similar such device. The first recommendation would be to combine a walk-man type tape player and a control device into one complete unit. This would produce a more compact unit. This would eliminate mechanical issues concerning the positioning of the two units. Due to the industrial type construction of the Libraries tape player the function keys require a large amount of force to be activated, thus requiring larger more expensive solenoids, solid state relays and increased power requirements. The second recommendation would be to re-evaluate the voice recognition chip used as this technology is evolving quickly and a chip that provided better background filtering might eliminate the issue concerning stopping the tape player when the volume is increased.