Voice apps spreading as standards mature

Voice technology is making its way into more mainstream uses thanks to standards-based technologies displacing expensive, proprietary platforms.

One of those applications is Rex, a disposable prescription bottle with embedded text-to-speech technology that can read medication details in a computer-generated voice at the push of a button. Tony Mariano, co-founder of MedivoxRx Technologies, came up with the idea for Rex when a visually impaired friend wound up in the hospital after mistakenly taking the wrong medicine.

Mariano is among those scheduled to speak at this week’s SpeechTek conference in New York. The semiannual SpeechTek trade show focuses on products and services based on voice technologies. About 3,000 attendees and 106 exhibitors are expected at the show, according to event organizers.

SpeechTek organizers developed conference tracks for key industries, including financial services, healthcare, insurance, government and utilities.

Speech vendors, too, have honed their wares for individual market sectors to improve adoption. Edify is launching an auto-attendant application for banks that automates typical customer requests such as changing a PIN and transferring funds.

Similarly, VoiceGenie Technologies will debut a suite of speech applications for utility companies. Its NXP Utilities Suite includes an automated meter-reporting system that lets subscribers phone in their service meter readings, and an emergency notification system that lets utility companies automatically deliver outbound calls to customers via a telephony or Web interface.

The trend to develop speech wares geared for specific vertical markets is a sign that the industry is maturing, says Daniel Hong, voice business analyst at Datamonitor. After gaining experience with general speech technologies, such as auto-attendant applications, vendors now are working to translate their expertise into vertical-specific products, Hong says.

But there’s still a lot of work to do. “In general, market awareness of speech is poor,” Hong says. Up to 40% of the largest U.S. companies still have only one speech application deployed, he says. Smaller companies have little to no knowledge of speech technologies.

That’s slowly changing. Improved IT spending among business is driving interest in voice technology deployments, Hong says. In addition, the advent of standards such as VoiceXML, and to a lesser extent the Microsoft-based Speech Application Language Tags specification, have freed companies from getting locked into expensive, proprietary speech platforms and custom-developed applications. “Across the board, pricing has declined,” Hong says.

Reaching the masses

These lower prices are expected to spur voice-enabled deployments. Adoption of speech recognition technologies is on the rise, according to Gartner. The research firm says shipments of speech-recognition telephony software are expected to triple between 2003 and 2008.

Six months ago at the spring SpeechTek conference in San Francisco, the big news was the debut of Microsoft’s Speech Server 2004 products. The launch marked the company’s entry into the server-based speech-recognition market, where it competes with vendors such as Nuance Communications, ScanSoft and IBM.

At the time, Microsoft touted the simplicity of its platform: Developers can use familiar Microsoft tools such as Visual Studio .Net to build voice-enabled applications, and Speech Server runs just like any other Microsoft server product.

Now applications based on Speech Server are beginning to debut. At this week’s SpeechTek show, Microsoft partner Pronexus is scheduled to unveil its VeoConnect speech-enabled auto attendant designed for Speech Server 2004. VeoConnect lets callers reach any department or employee by simply using their voice instead of dialing by name.

Pronexus is targeting small and midsize businesses that traditionally have been priced out of high-end voice-automation systems.

Vocomo also is aiming for affordability with its latest interactive voice-response platform. Designed for converged networks, Vocomo’s new VocomoVoice Response for VoIP lets companies deploy customer service applications in a single, IP-centric network without investing in specialized hardware, the vendor says.

It’s true that small and midsize companies have the ability to invest in speech, Hong says. “It may be on their radar now more so than ever before,” he says. But on the radar is not the same as on the budget. It still could take some time before the buying begins, he says.

Caller ID

Speech isn’t just for customer-service applications; it also can play a role in enterprise security. A pair of vendors are launching new voice authentication wares at SpeechTek.

Vocent’s DecisionMaker 2.0 combines voice biometrics and data analysis to verify caller identities before letting them create or access a bank account, for example. The new version includes enhanced reporting tools for auditing and logging purposes, and a refined risk-modeling engine designed to fight identity theft.

Similarly, Diaphonics uses voice authentication to help companies fight fraud. Its Spike Server confirms the identity of callers, records voice transactions and creates a searchable audit trail of all conversations. New in Version 1.5 is multi-tenant support so customers can provision multiple applications using a single Spike server.

While voice authentication also is getting attention, growth hasn’t been explosive, Hong says. Many financial institutions are at the early stage of evaluating and piloting the technology. But they don’t want to rush to deploy a technology that lengthens the duration of a transaction and potentially could alienate customers, he says.

Once the early adopters work out the kinks of how the technology fits into telephone transaction processes, , implementations will pick up, Hong says. “It will happen in a year or so.”