IVS, Inc. Announces Second Quarter Launch of Avstar
26 March 1998
The World's First GPS-Equipped Hands-Free, Eyes-Free, Transportable Interactive Voice Navigation Device for Cars and Trucks Monrovia, CA -- February 19, 1998 -- IVS, Inc., developer of interactive voice technology-based products for the automotive industry, announces that its premiere product, Avstar, will launch in the second quarter of 1998. As the world's first GPS-equipped hands-free, eyes-free, transportable interactive voice navigation device for cars and trucks, Avstar is poised to lead the vehicle navigation market in affordability, portability and safety. "Avstar is the only navigation product on the market that is entirely voice-interactive, allowing the driver's hands to remain on the wheel and the driver's eyes to remain on the road at all times," says Bob Evans, IVS, Inc. CEO and President. "Additionally, Avstar's affordable price point transforms the vehicle navigation device from a luxury item to a mass market product." Avstar navigation does not rely on display screens that require the driver to take his/her eyes off the road, but functions like a true co-pilot sitting in the passenger's seat. Powered by the vehicle's battery through the cigarette lighter and guided by a CD-ROM map, Avstar is easy to use. The driver tells Avstar his/her location and destination and Avstar determines the best route, then upon demand, gives verbal instructions turn by turn. Origins and destinations may be given as an address, an intersection or a landmark. Thousands of landmark locations may be given, including hotels, shopping malls, rental car agencies, gas stations, banks, restaurants, or just about any point of interest. Avstar does not respond with synthesized speech but rather, a human voice, that of voice-over artist Roy Lunel, who recorded all the street names, landmarks and points of interest. Although Avstar can be enhanced with an optional GPS antenna and receiver, it does not require one to function, unlike other vehicle navigation products on the market. When equipped with GPS, Avstar works even faster and smarter. The GPS model can automatically determine the user's starting point and provide warning signals for next-turns and the destination. While other navigation devices on the market must be either built in to the car or programmed into a laptop computer, Avstar is self-contained, lightweight and fully transportable, giving the owner the flexibility to permanently mount the unit or to transfer the unit to any vehicle. IVS, Inc. was founded in July 1997 as a result of a joint venture between Amerigon and Yazaki Corporation. Amerigon first developed radar-based backup and blind spot sensors, variable temperature seating, and aluminum chassis for electric passenger vehicles. Yazaki Corporation of Japan is a leading manufacturer of OEM automotive connectors and harnesses and supplies a significant percentage of the connectors and harnesses found in all automobiles manufactured worldwide. Based in Monrovia, CA, IVS, Inc. is an innovative company positioned to become a pioneer in the field of interactive voice technology. Staffed with many talented engineers and software specialists, IVS, Inc. is on the leading edge of interactive voice technology. IVS, Inc. is committed to developing products that will contribute to improving the quality of life through a full range of applications including voice-activated vehicle controls, communications, education, and entertainment. Engineered by IVS, Inc., Avstar is the world's first hands-free, eyes-free portable interactive voice navigation system for cars and trucks. How does Avstar work? Avstar is completely portable and plugs into the vehicle's cigarette lighter. Avstar may also be easily mounted inside the car. Through a microphone, the user tells Avstar his/her location and destination and Avstar determines the best route, then gives verbal instructions upon demand. Origins and destinations may be given as an address, an intersection or a point of interest. Hundreds of points of interest locations may be given, including hotels, shopping malls, rental car agencies, gas stations, ATMs and restaurants. Avstar even memorizes in its address book the user's own entries such as home, office or favorite night spot. What's inside Avstar? Two computers, some memory, software that understands the user's voice and converts Avstar's digital voice back to speech, and most significantly, the magical configuration to put it all together. The master computer runs the show, reads and interprets map information, and calculates routes. The voice processing computer listens to voice commands and responds to the user with spoken prompts. Why doesn't Avstar sound like a computerized voice? Because it doesn't use synthesized speech. A real person, Roy Lunel, recorded all the street names, and points of interest. That's quite an undertaking considering there are more than 50,000 street names in the Los Angeles area alone. Nationwide there are hundreds of thousands. How does IVS fit that many street names and landmarks on a single CD? The voice recordings are compressed by about a factor of ten and this compressed version is what is stored on the CD. When Avstar speaks, some tricky math formulas are used to restore the original sound. This particular compression/decompression algorithm is called GSM, originally developed in France for mobile telephone conversations. How does Avstar understand what the user says? By converting the user's voice into digital format, Avstar finds the patterns that are present in the word. Avstar is smart enough to ignore the loudness, speed and pitch of the voice because those properties don't change the word's meaning. Once Avstar has the patterns of the word, it compares them to patterns for all the words which make sense in response to one of its questions. Avstar picks the response that is most like what the user said. Can Avstar understand anybody's voice? Just about anybody. Avstar was trained to understand American English. Samples of how people talk were collected from about 350 people, young and old, men and women, Southerners, New Englanders, New Yorkers, and so forth. All those samples were analyzed to find the common characteristics of each word. Why does the user need to spell out the street names to Avstar? Because street names by themselves have no context. While some speech recognizers can understand large vocabularies, they get a lot of help from context. For example, if the user says "my dog eats beef," the recognizer knows enough grammar to expect that a noun will follow the verb. Since the verb is "eats," the object of the verb should be a food. All of this additional information reduces the number of words which are possible from tens of thousands to a much smaller number. No such grammatical or contextual help is available to Avstar. Any street name is equally likely regardless of what information has been given before. Recognizers also work on very large word data bases and are supported by many megabytes of memory. Avstar operates on less memory in order to keep the price affordable. But even if one had no limit on the amount of money to spend for recognition, it is still not possible to recognize any street names with no context. The technology simply does not exist today. Spelling introduces some problems of its own. Certain letters are hard recognize. Longer words contain more patterns, but letters are so short they don't contain much information. It's easy to understand how the recognizer could confuse B, C, and P because they sound alike.=20 Avstar anticipates the kinds of errors the recognizer might make and has a special spelling corrector to fix them. For example, you may spell "California" as "zjkysormyk" and Avstar will still figure out that you meant "California." This also allows the user to find the correct street even if they do not know the exact spelling. How does Avstar figure out the best route? Avstar examines all possible roads that lead out from the user's origin to the destination. Then computes the time it takes to travel those roads based on their length and posted speed limits. Finally, it selects the route that is quickest. How does IVS go about the mapping process? IVS has partnered with digital map suppliers Navtech and Etak.=20 IVS researchers pore over the street names to verify their accuracy and establish street names for roads that are not identified, such as freeway on-ramps. The maps are checked for authenticity against at least three other sources, such as U.S. geographical maps. The digitized map is transferred to a computer screen and displayed in text form so the roads may be recorded by the voice-over actor, Roy Lunel.=20 Posted speed limits on all streets are checked for accuracy and factored into the software. These are later used for determining the shortest routes. The next step is to run a series of routes, as many as 100,000 per map. The computer is programmed to pick a starting and ending point and print out the route in text form. Editors then examine the routes, comparing them against the actual map, to find the best routes and modify the digital map to correct bad routes. For a city such as Los Angeles, this process may take a month. For a city like Las Vegas, it may only take a few days. The maps are changed when warranted. What are the advantages of Avstar? Avstar is the only navigation product on the market that is entirely voice activated, allowing for the safest possible driving conditions. The driver never has to take his/her eyes off the road or hands off the steering wheel. Avstar is lightweight and portable, allowing the owner to take it from car to car and from city to city.=20 Avstar responds to queries with a human voice, like a co-pilot in the passenger's seat, not a synthesized computer voice. In addition to getting you from point A to point B, Avstar also recognizes many points of interest, such as ATMs, restaurants, hotels, shopping malls, rental car agencies, gas stations, and more. Because of the extensive computer testing and proprietary map-tuning techniques, Avstar calculates the best routes in the business. Avstar also calculates those routes faster than the competition. Avstar is the clear leader among vehicle navigation systems in portability, affordability and safety. Combine the interactive voice navigation technology with the upcoming entertainment and communications applications, IVS products promise to become essential tools for U.S.=20 drivers who spend more than 500 million hours per year in their cars. ### CD Maps are available in the following metro areas Atlanta Boston Chicago Dallas/Fort Worth Washington, D.C. - Baltimore Denver - Boulder Detroit Houston Indianapolis Las Vegas Metro Southern California Miami Minneapolis - St. Paul New Orleans New York City Tri-State Orlando Philadelphia Phoenix Portland - Salem San Diego San Francisco - Sacramento Seattle/Tacoma St. Louis Tampa - St. Petersburg
