Handicapped Projects Evolve Slowly by Peter Clarke Electronic Engineering Times, March 31, 1997, page 154 Electronic engineering has a tremendous scope to improve the lives of the handicapped. The fantastic possibilities of implanting an image sensor as an artificial eye, of directly stimulating the optical nerve, or of crafting a complete audio sensor to recreate hearing are being explored. There is even a project to implant a chip in the brain to capture experience and create "memories" - the so called "soul catcher project at British Telecom (see Feb. 3, page 1). But such efforts are at the leading edge of research. Early results from a few human subjects are telling researchers what is and is not possible, rather than providing tangible benefits to more than just a handful of individuals. Meanwhile, an enormous number of people around the world are handicapped in one way or in multiple ways. In Westernized societies, roughly 10 per. cent of the population over 50 years of age has impaired hearing. The figure rises to more than 50 percent above the age of 70. In the United States alone, There are approximately 4 million vision-impaired people unable to read even with magnifying devices. Where are the electronics to improve their lives? The answer is the same as it has been for many years. Though the industry is creating pacemakers, nearing aids and, more recently, novel applications and combinations of mainstream technology, progress is slower than it might be. Electronics thrives on a business model that produces high-cost, high-margin equipment for the few - such as the military and business concerns and gradually drives down costs and prices as the equipment comes out in higher volumes for the many: consumers. The telephone, radio, television, videocassette recorder and, more recently, the personal computer and mobile phone are good examples. But when the available market is by definition a minority, the equipment tends to remain expensive. For the majority of handicapped people especially those in the poorer countries of Africa and the Indian subcontinent - help in the form of electronic equipment is an unaffordable dream, just as telephone connections and PCs are. One of the first applications of transistor amplification was in a hearing aid. Yet despite the progress in digital electronics, until recently the hearing aid has remained essentially an analog instrument. Big-budget military research drove exploration of digital signal processing in the early 1980s. The consumer electronics industry picked up on that research in the late 1980s to create such things as compact disc audio systems and, now, digital television. Two hearing-aid equipment makers - Widex and Oticon, both based in Denmark - are now turning to digital signal processing for their top-of-the-range models. The standard in the industry is digital control of analog signal processing or all analog design. The Widex Senso, claiming the benefits of CD-quality sound and noise performance, was introduced in 1996 and is now shipping in volume. But it's not exactly a "first," said Peter Weis, a technical support engineer at Widex (Hellerup, Denmark). "The first digital hearing aid was made by a US company in 1989, but it was a large, body-worn device connected with a wire to the receiver (speaker) in the ear," Weis said. "For us, it was a question of when the technology would became available to allow us to implement a hearing aid without compromise." In defense of the hearing-aid industry, its application combines some extreme requirements that have delayed the use of digital technology. Miniaturization is necessary for behind the ear and in-canal models. Good battery life, another must have, has prompted the use of the zinc-air cell with a 1.4-V output that can drop down to about 1 V. Thus, mainstream CMOS technology at 5V or 3V, borrowed from the computer industry, has not been an option. "We had to develop our own BiCMOS technology at 0.7 micron geometry," said Weis. Senso has had a five or six year development cycle." It was in 1990 that we considered that digital technology would be available at about the time our digital control ideas would have matured. We then added other ideas, which we couldn't have done in analog circuitry." Widex achieved not only acceptable size and battery life performance for the Senso, but also a great deal of extra functionality. Rather than consuming more power than the previous generation analog model, it consumes slightly less. "And if we had tried to make Senso in analog, it would have consumed 100 times as much power," said Weis. The Senso DSP is a custom design tailored to a 20 bit data width and three frequency channel architecture, Much of the digital processing is used for flexible and programmable speech recognition and enhancement - an attempt to perform the difficult task of extracting sounds of interest, usually speech, from "noise." Other requirements are automatic gain control, feedback detection and prevention and noise protection. According to Lawrence Werth of London-based hearing-aid supplier and fitter PC Werth Ltd., hearing-aid costs in Europe range from about $800 to $2,500, including fitting. If a user requires a binaural fitting, it could cost as much as $5,000 to be set up with two of the Senso models. "One of the reasons hearing aids are expensive is because they need individual adaptation, fitting, the creation of ear molds and aftercare," said Weis. I don't think you'll ever be able to buy a hearing aid from a warehouse like you can a PC." However, he agreed that a self-fitted, low-cost hearing aid might make a worthy goal for the industry and help extend its penetration into the third world. Though Weis said he personally believes the future of hearing aids is digital, he declined to say whether his own company has abandoned analog research and development. "It might be that low cost analog hearing aids could coexist with digital designs," Weis said. Meanwhile, Ascent Technology Inc. (Boulder, CO), demonstrating the innovation that is a characteristic of the electronics industry, has come up with a reading aid for blind or extremely low vision users. The independent reading aid, or IRA, is a handheld scanner and enunciator attached to a PC that combines optical character recognition (OCR) with text-to-speech synthesis. It is intended to give blind users access to the print on everyday items such as mail, food packaging, newspapers, and medicine bottles. Such a device could be extremely valuable in allowing the sight impaired to live independent lives. Early development work was partially funded by the US National Institute on Aging and the Department of Education under a Small Business Innovation Research program. A first prototype went into trials during the summer of 1996. The user is guided to center the mouse on text using a vibro-tactile stimulator, and then to guide the built-in imaging array over the text. The PC reads the images under the mouse and "speaks" the words through built-in miniature speakers. By the fall of 1998, units based on a laptop Pentium-based computer, as well as custom designed wearable units, are expected to be in production. Initially the IRA will run in the $3,000 range, with an ultimate goal of $1,500, the company said. Ascent's strategy is based on leveraging technology developed for office automation, and it will be difficult to drive costs lower in the near term, said Jim Sears, president of Ascent Technology. "We're using an OCR in real-time, and that is why we need a swift processor such as the Pentium," he said. "Besides notebook based devices, we're looking at an embedded Pentium board with our own circuitry." Sears said the imaging chip used within the scanner, a CMOS device from VLSI Vision Ltd. (Edinburgh, Scotland), could potentially also include random logic circuitry. It is the sort of device that could help with cost reduction, he said: "That's the sort of technology that will help let this happen." In terms of driving costs below $1000, Sears said, "I'm sure we'll get there one day, but it's not obvious how. We'd probably need a neural network chip or something. It's a good question to ask every two years or so." Universal designs The real way to do it is, Sears continued, is through universal product design. You make a product that is useful to the whole population, which happens to be useful for handicapped people in one mode of operation." For example, he said, imagine the IRA with additional translation software. That would give the whole world access to English text in their native language and vice versa. It could be sold worldwide to drive down costs. It could make English accessible to an Ethiopian. The fact that it helps blind people would be a by-product." Text-to-speech synthesis is also being used at British Telecom Laboratories (Ipswich, England) to help 14 year old Holly Potter, who lost her voice through medical complications during an operation for croup when she was 3. She is now able to communicate using a notebook computer and program called Laureate, developed by BT, which converts to speech any text stored or keyed in by Holly. Unlike other computer voices, which are machine generated and sound artificial, the BT system produces real human speech using as its basic components recordings by another human being. While it is good to celebrate the electronic - engineering achievements that the past 25 years have brought to the few, it would even be better to be able to celebrate the extension of such benefits to people with the additional handicap of poverty. Perhaps that is the real challenge in developing electronics for the handicapped.