Peter Vescuso, Vice President of Market Development, Brooktrout Technology
NewsBlast: What role does Brooktrout play in providing speech technology solutions to clients?
Peter Vescuso: Brooktrout plays two significant roles in delivering speech solutions to clients. First, we've been a provider of telephony products, the primary network interface for speech systems, for over twenty years. Over that time we've developed a broad portfolio of telephony offerings that cover the spectrum of requirements for scale/density, network type (TDM, IP), etc. These products, which we've worked hard to optimize for speech applications, are an essential component of speech systems. Second, over that same period of time Brooktrout has invested in the support and sales infrastructure necessary to develop a world-class reseller channel that services enterprise customers. We believe that a well-trained and capable reseller channel is essential to the success of speech in the enterprise. As "pre-packaged" standards-based speech applications become a reality, our reseller channel can install and deploy these applications faster and at significantly lower cost than traditional proprietary speech systems.
NB: Discuss the Brooktrout roadmap for your future role in the 'speech solution stack' in 18 months or even a longer time frame?
PV: To understand our roadmap and what it means for speech solutions, it's important to explain where we fit in the solution stack and what we consider to be our core competencies. In terms of technology and products, Brooktrout's core competencies fall into two areas: media processing and call control. Both of these telephony-related technologies are essential to speech systems and usually are at the 'base' of the solution stack. Our products provide essential functions for speech systems by delivering speech efficiently to ASR systems (without noise, silence and echo) and by providing robust intelligent call control capabilities. Successful speech deployments depend on many things, and the telephony foundation is often times overlooked.
While we built the business delivering telephony (media processing and call control) products to developers and end-users in the form of board-based products, we continue to evolve our approach for delivering these essential elements. IP has become more pervasive and general-purpose processors have become more powerful. So we've extended our roadmap to provide developers with the same core technologies -- media processing and call control - but deliver them in the form of host-based software as well as board level products, plus provide a choice of IP, TDM, or hybrid network support. We believe giving developers the freedom of choice to deploy their systems using tried-and-true telephony boards, software-only, and a choice of connecting to any network infrastructure is the best approach to meet their needs. Going beyond just providing products with these features and capabilities, our strategy is to provide developers with a single seamless API that insulates their application from deployment choices related to hardware, software, and network interface. They can create a single code base, which simplifies development, distribution and support, and deploy any way they or their customers prefer.
NB: Please cite recent deployments where speech has enhanced customer service for your clients.
PV: We recently had a very interesting deployment with Vocomo for automation of clinical trials. Vocomo uses our TR1000™ speech processing board in their Voice Response speech-enabled IVR system. The customer in this case is Clinical Automation who uses Voice Response in their Recruitment Automation Service. The service allows physicians and researchers to recruit research subjects for clinical trials faster and more efficiently than with traditional screening methods.
In the past, potential subjects would read an ad in the newspaper, call a special number and then leave their name and phone number and wait for the researcher to call them back. Often researchers wasted valuable time responding to unqualified subjects. Now subjects can call a number and interact with the Recruitment Automation Service providing answers to questions via speech or touchtone. The service screens out unqualified subjects, freeing recruiters to concentrate their efforts on subjects with the greatest probability of enrollment.
NB: How are programming languages such as VoiceXML, SALT and MRCP impacting speech deployments? Are they important?
PV: All of these standards are very important and helping to eliminate barriers to speech adoption. Just as the first HTML web browsers provided internet access to mere mortals and caused the internet revolution, VoiceXML and SALT browsers provide a standard way to develop speech applications, and what's more important, they provide interoperability with existing web/data applications and leverage the extensive infrastructure of systems, developers, and expertise already in place. The interoperability with web/data applications that these standards provide removes what was a powerful barrier/impediment to adoption of speech. Unlike VoiceXML and SALT, which are effectively application-level APIs, MRCP represents a standard API for communicating with ASR/TTS resources. Using MRCP, a VoiceXML or SALT platform vendor can easily support any compliant speech engine; this provides freedom to choose the best-of-breed speech technology for a given application.
NB: Discuss some of your partners in delivering speech technologies. What are you looking for in selecting partners?
PV: One of our biggest initiatives is building our ecosystem of ISV applications, platforms, and resellers to develop the Enterprise market. We're looking for ISVs that have pre-packaged applications, or at least ones that can be deployed without weeks worth of installation and tuning, plus have the capability and support systems to be successful with a reseller channel.
In addition to application developers we're looking to partner with the leading speech platform providers. Last year we kicked off a very successful relationship with VoiceGenie. Very recently we announced relationships with Microsoft and Intervoice. Each of these partners offers a unique solution in the market and it's important for us to work with them.
ISV applications and speech platforms are necessary but still not sufficient to service enterprise customers appropriately. You need a world-class reseller channel. But the number of resellers we work with that are speech-savvy is insufficient today and we're looking to increase their ranks.
NB: How are you and your partners providing lower TCO and enhancing speech solutions for your clients?
PV: Given our position in the solution stack, you may not think there is much a provider of telephony products and technologies can do to impact TCO. But when you consider that all speech passes through our products on its way to the ASR system, it should come as no surprise that there's a lot we can do. For example, we've worked hard to enable our products to 'scrub' the audio signal to eliminate three big 'enemies' of speech systems, namely noise, silence and echo. "How does that lower TCO?" you may ask. The answer is -- and we have lab data to show this -- that by scrubbing the audio we can eliminate up to 30 percent of the processing a typical ASR client system would need to do, meaning less hardware is required, or perhaps the extra processing is used to deliver higher speech accuracy. All of these benefits either affect TCO directly or result in a better user experience.
NB: What are best practices an enterprise or service provider client should focus on when deploying speech?
PV: Our perspective on deployment comes from our place in the solution stack. We recommend that partners consider the performance of the entire system from the choice of ASR to the choice of telephony. We also believe it's important to choose and exploit best-of-breed components in standards-based systems - not all telephony products are the same. For example selecting telephony products that offer superior echo cancellation can avoid deployment surprises and extensive time spent "tuning" the system.
NB: What will be the biggest event this year in offering speech technologies solutions?
PV: The biggest industry event this year is the introduction of Microsoft's Speech Server. I think everyone realizes Microsoft is in the best position to drive speech into the mainstream of the IT market, which will be good for Brooktrout and the industry as a whole. With respect to Brooktrout, the biggest events for us this year are our partnerships with Microsoft and Intervoice, as well as our new line of TR1000 telephony boards and support for host-media processing.