Learning

Text-to-Speech Reads to the Blind, But What More Can Tech Do?

The author's father using the Stereotoner to read.

The author’s father using the Stereotoner to read.

Back in the 1970s, before a personal computer was on every desk or lap and a smartphone in every pocket, blind people read printed material—books, newspapers, bills—with reading machines. Harvey Lauer at the Blind Rehabilitation Center in Hines, Ill., was a pioneer in developing reading machines for the blind, and my father, Richard Bennett, a researcher at the Veterans Administration in Palo Alto, was one of his colleagues.

“Blindness,” Lauer once said, “is something more than a nuisance, but a lot less than a major catastrophe.” The phrase aptly sums up the challenge of reading for blind people: It takes effort, but it’s not an insurmountable problem.

Among the devices my father and Lauer worked with in the 70s were two that seemed to presage great things: the Optacon and the Stereotoner.

The Optacon used a small handheld scanner to convert text into tactile patterns. A blind reader would place a fingertip on a small bed of vibrating pins that rose and fell as the optical scanner moved across the text. The vibration had to be adjusted so as to not cause numbness in the finger, and the pin area was precisely designed: small enough to make it possible to read quickly with just a fingertip; large enough that the characters were discernable. The Optacon had several thousand users before production was discontinued.

The Stereotoner emitted tones while scanning text, based on the shape of letters and other typographical characters. Skilled users could read about 60 words a minute. Lauer said those with an aptitude for music—something he and my father shared—tended to fare better than others. The Stereotoner was nowhere near as widely used as the Optacon, despite being more portable, in part because of the steep learning curve. An earlier incarnation of the machine was called the Visotoner.

More sophisticated versions of these devices never came into being, and though both offered a valuable way to read some material some of the time, the approach has been surpassed by new and different technologies.

The author's father wearing a pair of sonar glasses.

The author’s father wearing a pair of sonar glasses.

I set out recently to learn how blind people are using today’s technology to read and navigate the world.

* * * *

The biggest disruption to the text-to-tactile and text-to-tone approach was the development of machines that scanned type and read back the text in a synthesized voice. Technology visionary Ray Kurzweil introduced one in the late 70s, and most of the devices that have followed use some kind of text-to-speech system.

Companies such as Freedom Scientific, Optelec, Adaptive Technology Resources, and HIMS offer products that use speech technology and allow blind people to perform a number of daily tasks: use computers, convert electronic documents to Braille that can be read on a handheld device with a moving band of type, or scan and save written pages so they can be  listened to on a range of portable and desktop devices.

Among the researchers following the path laid out by Harvey Lauer is Joshua Miele at the Smith-Kettlewell Eye Research Institute in San Francisco. Miele has leveraged new technologies to develop a host of blind accessibility tools.

Ray Kurzweil with his reading synthesizer.

Ray Kurzweil with his reading synthesizer.

Was the path laid out by the Optacon and Stereotoner a dead end? “No,” Miele says. “But cheaper, more convenient, and more scalable ways to read have supplanted them.” Much text is available in machine-readable format now–even restaurant menus can be found online and read back.

Still, Miele says the Optacon or Stereotoner, built more cost-effectively today, might find other applications. “The old machines are useful for object-specific use, such as reading a pill bottle or a map, trying to ascertain what magazine is in your hands, or reading graphs and charts,” Miele says.

Mara Mills, a professor of media, culture, and communications at New York University, says text-to-speech technology will continue to predominate. “Synthetic speech may not have been inevitable, but strong social forces advanced this output over text-to-tone or text-to-vibration,” she says. “To the engineers who worked on reading machines in the postwar period, speech and speech-like sounds seemed easier to decode than tonal or vibratory patterns. Speed became the guiding principle for reading-machine development.”

To be sure, reading “dog slow”—to quote Miele—with a Stereotoner may not compare to hearing an audio version of the latest novel or having a smartphone give you directions while walking around town.

Miele has faith in the potential of mobile devices. His lab’s WearaBraille was a proofofconcept project that let a blind person interact with an iPhone, Android smartphone, or Windows laptop by finger-tapping while wearing accelerometers embedded in a wi-fi glove.

The rise of consumer voice recognition products such as Apple’s Siri made the WearaBraille redundant. But Miele’s lab has also developed tactile maps for Bay Area Rapid Transit stations and other places, and a cloud-based, crowdsourced software program, the Descriptive Video Exchange, that lets people describe scenes in a video or movie so that a blind person can follow along. He’s also working on a platform that would turn the smartphone into a household master remote, to control wirelessly-enabled devices from a Roku video streaming device to a washing machine.

Second Sight, based in Sylmar, California, has developed an “implantable visual prosthetic” device called the Argus II that, according to the company website, “restores some functional vision” for blind people by allowing visual signals to be sent straight to the brain, bypassing damaged photoreceptors. The device uses a combination of glasses, a camera, an antenna, and video processing unit to “create the perception of patterns of light which patients can learn to interpret as visual patterns.”

In a similar vein, Nano Retina, based in Israel, is working on an “ultra small, easy to implant, artificial retina designed to restore sight” to blind people. According to the company’s website, the surgical process will take about 30 minutes, using local anesthesia, with return of sight expected to be instantaneous.

The idea of using glasses to aid in mobility is not new. In the late 70s, my father used a pair of sonar glasses which, when connected to a special cane, allowed a user to “see” objects in front of him while walking down the street. He demonstrated the device in a National Geographic movie segment.

Fast-forward almost 40 years and we have Google Glass, which Miele agrees “has potential” as a game changer for blind people. “It all depends on the level of accessibility built into the device. If it doesn’t have a standard framework for something like text-to-speech, or for creating apps, or for non-visually navigating an object, it will be harder for us to create accessible tools.” Miele points out that iPhones are accessible out of the box: “Blind people own them, and building tools on it makes sense.” He says the jury is still out on Google Glass.

* * * *

I have no doubt my father would have been amazed by these technological advances. Yet, some promise remains unfulfilled. Years ago Harvey Lauer, 79 and blind since birth, told me the key for him was being able to “get his nose in a book”—another way of saying there’s a difference between reading with a machine and being read to by a machine. Print is, after all, designed for the eyes.

The Optacon and Stereotoner were eyes of a sort, and their limitations would likely have been overcome eventually. Lauer still keeps them on hand for smaller tasks.

His ideal machine—the one he hoped would be created in the 80s when speech technology came on the scene—would have combined tactile, tonal, and speech options, permitting a blind reader to more fully experience the act of reading.

Mills agrees that approach would have value: “Where optical character recognition fails with non-text items on printed pages or with atypical and low-quality printed materials, direct translation to tones or vibration can fill in.”

Had Lauer’s ideal machine come to fruition decades ago, today’s reading devices might be different, he believes. “What’s important,” he says, “is that the technology is there as a record of what we did—and as a way to inspire those who came after us.”

Tags: , , , , , , ,