The first quarter of 2011 has brought two telematics launches, Audi of America and Hyundai Motor America, both eschewing a traditional call center. These two companies join Ford Motor Company on the call-center-less bandwagon. These decisions are momentous for telematics service providers.
The entire industry has a powerful interest in cutting the cord to expensive call centers to reduce costs. The question strategists must ask themselves is whether they can afford to make the sacrifice of human contact in both safety and mission-critical applications as well as for POI lookups and concierge-type functions. Is speech recognition technology up to this challenge?
On the subject of manned call centers, researchers do not appear to agree. Hyundai claimed in its presentation at CES that its research showed consumers preferred to converse with machines. Colleagues contacted at OnStar begged to differ, saying their research showed just the opposite. The “truth” lies somewhere between these two positions and may reflect demographic elements.
What research subjects are probably saying is that they will accept a machine-based voice recognition system if it works. Suppliers are hard at work trying to deliver such a solution including Vlingo (ATX), Tellme (Ford), Voicebox (Toyota) and Nuance, working with all of the above.
Vlingo, Tellme, Voicebox and Nuance, but no Google Voice
Curiously absent from this mix of server-side recognizers in the automotive industry are Google Voice and whatever voice solution Apple may have up its sleeve. With neither of these organizations – Google or Apple – directly targeting automotive solutions, it is unclear how their solutions stack up against those already deployed for automotive systems.
To help make the case for IVR-based systems, ATX presented its own research on the subject at the recent Fully Networked Car event at the Geneva Auto Salon in March. ATX concluded from its study – conducted with Virginia Tech – that server-based voice recognition was superior to manual task performance for a variety of functions and that users rapidly became accustomed to using voice commands and preferred the voice-based solution.
It is one thing for voice to outperform manual task performance. It is another thing for server-based speech recognition to outperform on-board speech recognition in an automotive environment.
Each of the existing server-based recognizers has experienced challenges. It is not clear whether the failures of existing systems are due to vocabulary size, syntax, architecture or echo cancellation/noise reduction technologies. More work clearly needs to be conducted.
What is clear is that voice is rapidly emerging as a critical market differentiator. Ford has already demonstrated the power of voice as a vehicle for defining its brand. ( http://bit.ly/c6IRfn - #Nuance, #Ford Lead New Automotive Branding Paradigm - Lanctot)
Ford has suffered some setbacks recently after greatly expanding its vocabulary and altering its recognition architecture. (http://bit.ly/eiUNul - MyFord Touch Provides Mixed Bag of Changes to Speech Interface – Schreiner) Consumer Reports magazine was quick to pounce on the changes which manifested in MyFord Touch. (http://bit.ly/hJNuGu - MyFord Touch: Compelling or Overwhelming? – Schreiner)
But there were significant recognition quality differences between Ford’s on-board and off-board recognition systems. Most, though not all, OEMs have their own in-house teams working on these challenges. Tier One suppliers, too, are tackling the issue, as are the voice suppliers themselves. The leaders that emerge will be those that are best able to match their understanding of human speech patterns and behavior with automotive grade systems.
Implications:
Much is at stake in getting server-based voice recognition to work properly. Hyundai may have opted for an IVR, but they also opted to back it up with call center personnel assisting with recognition in the background. Younger people are likely to be most receptive to IVR systems after getting their first exposure to voice recognition via Google Voice-assisted searches on Android phones. But voice recognition in a noisy moving vehicle is a more complex and risky proposition.
IVR systems do away with many of the evils of manned call centers including cost, wrong numbers, and prank calls. The industry shift away from call centers puts the pressure on those organizations still offering manned solutions to justify those investments with system enhancements – including superior integration of vehicle and customer data.
But the shift to IVR also reflects the industry trend to limit data access rather than fostering it. The customer may not care about getting personalized attention from a human being as long as they can get the information they need about their car or the assistance they require in an emergency.
The dominant call centers in the automotive industry are currently associated with embedded telematics systems (OnStar, ATX), emergency response (Intrado, ADT) or roadside assistance or insurance (AAA, Road America, Cross Country Automotive Services). But there are many other call centers, particularly those of the wireless carriers, that are in position to challenge the automotive incumbents. Even as the auto industry call centers gear up for an IVR future they will have to enhance and maintain their automotive grade solutions to hold off a potential wireless industry challenge.
Most important of all, though, will be managing the transition – as ATX is doing carefully with Hyundai – to ensure that IVR systems are not deployed before they are customer ready. Expect a lot more testing and research in the coming months and years.