Honda and IBM Deploy Advanced Embedded Speech Technology in Cars
Honda and IBM made a joint announcement about the deployment of IBM's embedded speech technology in two Honda high-end vehicles being sold in the United States and Canada. The speech system comes standard in 2005 Acura RLs and it's an option for 2005 Odysseys.
This is the third year that Honda has deployed IBM speech in its automobiles. Each year the system has increased in scope and functionality. This year, voice-activated command-and-control functions include an expanded set of commands for the radio, audio, and climate-control systems. Each of which includes a number of variations so that drivers aren't forced to memorize command structure and wording.
The most significant enhancement is that embedded, in-vehicle ASR and TTS have been built into the vehicle's navigation system. "It's completely integrated with Honda's navigation system" explains Alistair Rennie, vice president of sales and marketing of IBM's Pervasive Computing division (PvC). "And it's the first in-car navigation system to offer speech recognition to identify street and city names across the United States." This means that a driver can say any of the 1.7 million city and street names in the system with the full expectation that the system will be able to recognize and use them to generate turn-by-turn driving instructions. According to Rennie, "This ability to recognize street names is a huge advance in recognition quality."
Navigation instructions that are given to the driver are a blend of concatenated TTS and pre-recorded elements (e.g., "turn right") using the same voice. Barbara Britt, program director of the PvC Embedded Voice Department describes it as "a smaller system than the one we use on a server but it's the same technology. We recorded many, many hours of speech and integrated the TTS with the existing recordings made by that same speaker." The blend of TTS with pre-recorded segments produces natural-sounding speech. Britt adds, "This means that it doesn't simply say 'turn at the next intersection.' It will say the name of the street where you need to turn."
The navigation system is integrated with Zagat restaurant guide and information. "You can request directions and names of nearby restaurants by category" says Rennie. "You can even have the restaurant reviews spoken to you or displayed visually on a screen. Then, you have the ability to determine whether you want directions to the restaurant or not."
These in-vehicle capabilities are enabled by a set of DVDs that are provided with the system. Cars sold in the United States come with a set of DVDs that cover the forty-eight states of the US. Vehicles sold in Canada contain DVDs for all Canadian provinces. The Canadian speech system is available in English and Canadian French.
In addition to these embedded in-car operations is HandsFreeLink, a Bluetooth voice-dialing system developed for Honda by Johnson Controls. It provides a wireless interface with mobile phones that allows calls to be placed and received via the car's embedded speech recognition and audio interfaces. In 2004, HandsFreeLink became the first hands-free Bluetooth phone system offered as standard equipment in an automobile.
Honda has pursued in-car speech very aggressively. "It's one of the fundamental ways they're differentiating themselves from their competitors" says Rennie. The speech system also serves two other objectives that are important to Honda: safety and customer satisfaction. Britt explains that "safety was the primary motive for using the speech. Honda was not only looking for ways to improve their navigation system - which is one of the best in the world - but also to provide additional safety. They wanted to include speech so that the driver didn't have to keep looking down at the screen to interact with the system."
Customer satisfaction is equally important to Honda. Rennie describes this as "wanting to bring joy to their customers." This means that the speech functions and those of other software systems are designed to be seamless - to operate in the background so that the driver can enjoy the driving experience.
Calling speech the "sweet spot" of telematics Rennie envisions a time when it will be part of a "server managed client" configuration in which you can remotely manage the car from a server. "It becomes a no-touch model. It's the ability to proactively monitor and upgrade software without the involvement of the consumer or a technician. That's the goal."
Rennie summed up IBM's attitude toward yesterday's joint announcement and IBM's long-standing relationship with Honda. "We are very proud of the work that our team has done working very closely with Honda and Honda's partners to create what we think is the most advanced in-car speech recognition system on the market today. It's a major step forward for Honda and a great day for IBM as a technology provider. Plus, it's pretty exciting news for the automotive industry and the speech industry as well. It underlines the fact that speech is becoming a mainstream, consumer-friendly, and seamless part of a lot of environments - not just automobiles."