The History of Video Voice

Video Voice has a more than 25 year history. It got its start back in 1984, when a team of scientists, one of whom was hearing-impaired, began looking for a way to illustrate speech. The chemistry professor, who had been deaf since birth, decided as a child that he wanted to be a scientist, and understood that he’d need to be able to speak to be successful in that career. So he regularly attended speech therapy and diligently practiced at home between sessions, but was often frustrated to find he’d been practicing the wrong sound. So he and colleagues - an electronic engineer and a software designer - set about creating an electronic device that would illustrate sounds as they were produced. Their design was based on the vowel representation scheme from Grant Fairbanks’ Voice and Articulation Drill Book (©1954), and they  were granted a patent for both the hardware and software.

The first version of Video Voice was based on a small microcomputer called the Interact. Most people haven’t heard of it, but it was one of the first personal computers on the market. It was released about the time of the Radio Shack TRS80. The graphics capabilities were modest, with only a 112x77 pixel resolution (!!) and a total of 8 colors, but it had a built-in analog-to-digital converter, an important capability since speech is analog by nature. The earliest Video Voice models included that computer and an external device called the Speech Analyzer (or "black box") that converted the voice data to digital form as it was sampled.

The inventors’ prototype provided feedback that was pretty meager, nothing more than a few dots on the screen that showed the basic location of a vowel sound. That didn’t seem like it would be interesting for long, so we set about fleshing out the display and software to turn it into a tool that would be motivating - adding color, a model and trial structure, on-screen vowel display, and much more.

As computers gained popularity in schools, therapists started asking for a version that would operate on Apple II/IIgs computers, so we converted the software to operate on those platforms. Then came the Macintosh, and we produced a Mac-based Video Voice. And then one for the IBM PC, first a DOS-based version, then a Windows-compatible one. (During this time, IBM produced its Speech Viewer program, which became widely known, but is no longer available.)

The external Speech Analyzer was retired with the release of Version 3.0. All voice sampling is done through the computer’s internal sound capabilities, and the analysis with our own specialized software routines. This allowed us to greatly expand Video Voice’s capabilities to increase the frequency ranges of sound sampling and add many new games and displays, at a signficantly lower price,.

Expanded capabilities include much wider pitch range to accommodate low-pitched male voices and high-pitched children’s voices (something the Speech Analyzer versions were limited in). We’ve also been able to increase the formant frequency sampling to illustrate and differentiate high frequency sounds like /s/ and /sh/. (The earlier, hardware-dependent versions could detect the presence or absence of high-frequency sounds, but could not tell the difference between them.)

With faster computers with greater capabilities, we’ve been able to greatly enhance the graphics used in Video Voice. They’re still not as fancy or with Xbox-type resolution, because there’s a lot going on "behind the scenes" in the voice sampling and analysis that takes substantial "compute time." And, after all, the point is to illustrate speech, not to be a realistic action game.

To wit, many years back, some folks designed a software interface that integrated with some video games that were then available which had higher resolution graphics and action. It could be programmed to accept 4 words that would control the action of a game (for purposes of this example - "left," "right," "up," and "down"). The goal was to command virtual game player to move around and avoid being attacked by a monster (again, an example). Unfortunately, what hadn’t been considered was the excitement factor in the sound analysis. Targets that were calmly produced when the game was being initially set up, didn’t achieve the desired motion response when the player got excited during the game action and began shouting the words at the screen. Pitch and volume, after all, do affect sound production! This program quickly faded from the scene.

Version 3.0 is the only Video Voice model now being produced. It operates on most Windows operating systems (Windows 2000 and later), and is not dependent on processor speed. In fact, on really fast computers, we actually have to slow some things down. A two-second model, for example, needs to be two-seconds long, even if the computer is capable of displaying the graphics much faster.

That’s the basic history of Video Voice. Development is ongoing, with new things added all the time, so there will still be future chapters written!

Video Voice Support Team
1-800-537-2182
mv@videovoice.com