Speech Technology Magazine

 

First Let’s Kill All the Hardware

At a recent speech conference—Nuance’s excellent V-World—I was again struck by the combined power of speech recognition, voice-user interface (VUI) design, TTS and strong back-end integration for content delivery. At the risk of sounding fatuous, quality has become so much better that the applications are fooling people into treating recognizers like, well, other people.
By Mark Plakias - Posted Jul 11, 2002
Page1 of 1
Bookmark and Share
At a recent speech conference—Nuance’s excellent V-World—I was again struck by the combined power of speech recognition, voice-user interface (VUI) design, TTS and strong back-end integration for content delivery. At the risk of sounding fatuous, quality has become so much better that the applications are fooling people into treating recognizers like, well, other people. Please note the paragraph above contains no mention of high-density, supercharged hardware platforms, connecting callers to these resources. In a recent conversation with a client, the frustration with the lack of adoption of speech solutions by enterprise customers was almost palpable. With all that is now possible—and the clear benefits of finally being able to align the phone channel with the rest of enterprise business processes via XML-based component technologies—the client is asking: What’s holding up the show? The usual suspects are (1) the cost of speech recognition software licenses and (2) the cost of hardware. While we can debate the right of speech-recognition vendors to make a living, we’re not going to do that here. On the other hand, one of the things that is increasingly controllable is hardware—the gateway, the telephony piece, whatever you want to call the boxes that terminate the T1/E1/PRI or other flavor of network connection. You know, the boxes with all those blinking lights. Historically these have been specially designed because there is a premium to be exacted for guaranteeing performance in high-speed communications applications. But what is now different, in addition to more reliable off-the-shelf (OTS) platforms for handling inbound calls, is the ability to separate those boxes from the software components that create the magic of the voice application. Magic equals value; hardware equals cost and headaches—nothing complicated here. If you don’t like the cost of the hardware and you think that’s a problem for your customer (in terms of committing to a voice solution), get rid of it. It’s that simple. This is not a plug for third-party hosting providers or wireline carriers who “get it” (a list that is growing longer every day). But we recommend that solutions providers expand their channel strategies to include hosting partners that can take the platform cost off the table, either completely or partly. Major enterprise customers such as Merrill Lynch and Delta are now seriously adopting hosted solutions. And on the channel side, consider the adoption by Call Interactive of the Telera platform. On the one hand is a company that’s been doing high-volume outsourced IVR for more than a decade (Call Interactive), on the other hand is a software developer (Telera) with a “borderless” solution that can run as customer premises equipment (CPE) and spill over onto hosted facilities for overflow traffic—capping hardware costs. But the big channel story is telco. Carriers don’t have the option of avoiding capex on platforms. Service providers, and specifically wireline and wireless carriers, must invest in voice platforms to do business. Whether it’s VAD, DA, voice messaging or business office operations, the cost reduction pressures in the telco industry are propelling over 125 deployments already. The point is simple, there will be excess capacity on these platforms. Carrier-hosting initiatives are already underway at BellSouth, BT, NTT, Qwest and Worldcom, and virtually every carrier with any kind of toll-free business is looking very hard at this. The Kelsey Group is projecting $2 billion in carrier hosting and voice-enabled toll-free revenues by 2006. That, in turn, is a supported by a whole lot of blinking lights. That’s good business for high-density voice platform suppliers. It’s also good business for solutions developers that have customers balking at buying their own platforms. A final thought: If you want that box on your premises, but you don’t want to pay for hardware, pay for transactions instead. Per-call or per-transaction pricing from hardware manufacturers isn’t all that new. Platform providers have always flirted with managed service scenarios and pay-as-you-go pricing. That’s not unique to voice recognition, but voice is a very good candidate for these kinds of pricing scenarios. You don’t know if you don’t ask, right? Mark Plakias is a senior vice president and managing director of communications and infrastructure for The Kelsey Group. He can be reached at (609) 921-7200.
Page1 of 1