Keith Byerly, Product Manager, Brooktrout Technology

Q Tell us a little about Brooktrout and where do you expect to be in three to five years?

A Brooktrout Technology makes products for voice, fax and call processing used in the development of applications and services for the New Network(TM) -- a network born through the marriage of the PSTN and IP networks.

The company was founded in 1984, and has been a publicly traded company since 1992. Its 350 employees are located around the world, with offices throughout the U.S., and in Belgium, Canada and the United Kingdom.

The company's overall strategy is two-fold: deliver technically superior board-level products for the enterprise and service provider markets, targeted at the unique and rapidly evolving requirements of speech and IP-based systems; and collaborate with its partners to help them increase existing business, expand into new markets and accelerate the delivery of new value-added applications and services. Brooktrout partners with other leading technology vendors to ensure maximum interoperability and performance at the application and system level.

Brooktrout believes that the impending maturity of the speech and IP segments will result in significant consolidation within our industry during the next three to five years. We're already seeing our competitors abandon the board-level product space, offering system-level platforms in hopes of higher margins. Microsoft's recent entry into the speech market signifies a shift from an early adopter to an early mainstream market, with all the competitive issues and opportunities that represents. Brooktrout is exceptionally well positioned to thrive in the standards-based, technology-driven environment of the future.

Q How is Brooktrout improving speech technology for your customers?

A Brooktrout was not the first to market with voice boards, and we recognized that our products had to deliver significant performance benefits if we were to be successful. One of the problems we saw is that speech systems still suffered from a number of performance bottlenecks that we could address with appropriate use of our DSP and processor technology. It also became clear that our competitors were continuing to produce "just another telephony card."

So, we led the industry with improved echo-cancellation technology to improve recognition accuracy during barge-in. We implemented advanced voice activity detection to provide first-pass endpointing compatible with speech systems in order to boost system scalability. We optimized our drivers and firmware, then demonstrated the highest-density speech system on the market. Going forward, Brooktrout will continue to invest in providing superior enabling technology to optimize overall speech system performance, scalability and cost.

Just as important as this advanced technology is what we do to make it easily accessible to our customers and partners. Although it may come across as trite, Brooktrout's "True Partner" philosophy is real, and at the core of our company values. We invest in our customers' success by selling products direct, providing free technical support and training from the factory, offering joint marketing and sales programs, and collaborating throughout the development cycle to get them to market - and to revenue - as quickly as possible.

Q What are your enterprise customers telling you about utilizing speech in their communications platforms? How about your customers supporting the service provider space?

A On the enterprise side, we see two trends. First speech access is driving significant new demand in the otherwise stagnant IVR segment. Proprietary IVR systems have begun to move towards an open systems architecture, leveraging off-the-shelf hardware technology and state-of-the-art speech software to deliver value-added capabilities more quickly and at a lower cost. Although this investment in speech was initially justified by potential cost savings, demonstrable gains in customer satisfaction and call completion rates are fueling demand. The second trend is the "opening up" of traditional proprietary voice messaging platforms. Developers of these systems are supporting IP and VoiceXML and allowing direct integration with corporate IT networks. We believe this will spawn a new round of innovation for corporate messaging applications.

Not surprisingly, the service provider segment has been slower to invest in new technologies. They must balance the demand for new value-added services and revenue streams with limited capital budgets and conservative technology adoption. However, enabling technology providers such as Brooktrout are now seeing development starts accelerate rapidly as this segment prepares for the end of the current slump.

Q What are some issues that will impact the deployment of speech applications over the next few years and how should the industry address those issues?

A A pervasive issue has been the high cost of high performance. Computing technology has finally become just powerful enough to support speech systems with the accuracy expected by consumers. To reduce the cost of deploying a speech system, developers will need to support higher density speech processing through faster processors, advanced voice processing resources (such as DSPs), or both.

Of course, additional computational power must also be applied to the problem of deploying reliable speech systems. The extensive tuning needed to ensure robust performance adds significant time, cost and risk to the financial equation today. Although improved optimization tools and experienced professional services organizations can help, only time (and Moore's Law) will truly solve this problem.

Finally, as the market and technology continues to mature, a variety of standards will be proposed and eventually adopted to ease the design and deployment of speech systems. We're seeing that now at the application level, with VoiceXML, SALT and their derivatives rapidly evolving; Microsoft's entry is sure to drive the API standards process forward at an accelerated rate. Speech technology and telephony platform vendors continue to offer proprietary APIs, reflecting the relative immaturity of their technology and markets; as a result, third-party VoiceXML platform vendors have emerged to hide this complexity from the application developer. Speech will become truly pervasive when open standards are adopted throughout the entire solution stack, driving down costs and promoting innovation at the application layer.

Q What are your thoughts concerning multimodality and its potential impact on communications?

A There's not much doubt that, although speech is the natural choice for relatively simple input, selection and commands, it has severe limitations as an output device. Entire classes of very useful interactive applications depend on the efficient presentation of complex information. Multimodal capabilities complement the natural interactivity of speech to increase the scope of tasks that can be performed in a mobile, distributed computing environment. The communications network of the future must provide "what you want, where and when you want it," and multimodality will play a key role.

Q What would you like to see from the speech developers?

A From my admittedly biased perspective, I'd like speech developers to consider how the accuracy, scalability and cost of the overall system can impact the success of their application. Of course, many speech developers would prefer to think of telephony boards as commodities (or not think of them at all), so I'd also like to see them using standardized APIs, tools and platforms that utilize best-of-breed technology - like that from Brooktrout.

Q What are recent deployments that you believe highlighted the value of speech?

A Several recent announcements, although not the most significant from a revenue point of view, highlight the value speech is providing across a spectrum of applications.

SpeechWorks' collaboration with Ford Research for the design and implementation of a multimodal interface will enable car occupants to operate on-board systems including entertainment, navigation, cellular telephone and climate control by voice. This partnership signifies that speech has finally achieved the credibility needed to move into the mainstream auto market. It also demonstrates the accepted value of a multimodal interface, especially for complex applications such as navigation, delivered within the constraints of an embedded speech platform.

Nuance's new voice-activated 511 phone service in the state of Washington will allow drivers to get information on current weather, congestion and road conditions by voice. In a very public and visible application, speech technology has improved the value and usability of the system by integrating a wide variety of information into a natural dialog that callers can quickly navigate.

Q Who are some of your partners and how did you choose these companies? How can someone become a partner with Brooktrout?

A Brooktrout partners with leading speech technology and application vendors to provide our mutual customers with best-of-breed solutions. Our partners range from well-known companies such as SpeechWorks, Nuance, Philips and Lucent to less familiar names such as Rhetorical Systems, Loquendo, Cambridge VoiceTech and Closest Point Solutions.

Our most successful partnerships are based not only on complementary products and services, but also on shared values of close partnership and technical innovation. Prospective partners are encouraged to learn more about our Partner Access Network Program™ (http://www.brooktrout.com/accessnetwork/), which provides early access to our newest technology, direct technical and porting assistance, as well as joint marketing opportunities and lead referral programs.

Q What vertical markets do you see increasing their buying of speech applications?

A As an enabling technology provider, Brooktrout does not sell directly into vertical markets, so I can't provide a personal perspective. Based on discussions with our customers and partners, the financial and mobile sales markets continue to show demand based on a demonstrable ROI.