November 1999

[Speech Technology Efforts Talk To Developers] [Not “Just” Browsing ]

Speech Technology Efforts Talk To Developers

When I was in college, I volunteered for several projects within my university’s linguistics department. For three hours a week, I listened to computer-generated sounds and determined if they sounded like a “bee” or “dee,” an “em” or “en,” and so forth. The group’s research was the early stage of development for a speech application. Offering my services to the linguistics group is the extent of my experience assisting in the development of speech technology; however it is enough for me to realize the importance of potential applications.

I am not alone with my opinion. The future for speech technology is looking bright. Several strides forward in development of speech applications have debuted in the past few months. One of the latest is a joint effort by Unisys and Microsoft, who are working together to accelerate the adoption of the Speech Application Programming Interface (SAPI) for speech applications, providing software developers with a rich set of tools and support programs that will make speech-based technology easier to deploy.

The companies’ objectives will be furthered by the launch of SpeechDepot, a Web site that provides the components, information, expertise, and products needed for design, development, and deployment of speech-based applications. The site contains development tips, including slideshows and excerpts from seminars; links to other speech technology sites; and program updates and patches for download. SpeechDepot is meant to serve as an electronic forum for the speech application development community and provide a place where they can obtain, share, and disseminate information on speech.

As part of this new initiative with Microsoft, Unisys has created a specially packaged SAPI-only version of its Natural Language Speech Assistant (NLSA) application development toolkit, which is available for download on the SpeechDepot site. NLSA is a toolkit that streamlines voice-user interface design, development, and testing at deployment, cutting the average time for creating a new application from months to days.

Meanwhile, Aculab has worked on broadening the appeal of its Prosody speech and signal-processing platform, which provides developers with versatile, high-density speech processing in a compact and flexible form. The latest Prosody card configurations have been released to assist the formulation of entry level and lower cost developments in a lab environment, through to the release of full-scale solutions. The development compliments the product portfolio that can offer either independent digital access and speech resources or powerful combinations of the two.

The Prosody PCI range can now provide, as a minimum, 30 ISDN lines and 60 speech channels via a single DSP, through to a market leading 120 ISDN lines and 240 channels of speech. A combination of 1, 2, 3, or 4 floating SHARC DSPs as well as 1, 2, or 4 digital trunk connections can now be selected depending on a particular project’s requirements. The completion of the range opens Prosody’s appeal to wide market, who will be able to enjoy a selection of Aculab’s free speech algorithms — including Text-To-Speech, Automatic Speech Recognition, fax, DTMF pulse detection, and conferencing — and protocol support in a single slot solution.

Recent progress in speech technology comes from Phonetic Systems with version 3.0 of their DirectoryAssistant. Every day, telephone companies have hundreds of thousands of updates to directory assistance listings. DirectoryAssistant makes these updates immediately available for speech access without any special speech tuning or human intervention, thereby reducing the long-term costs of ownership for the providers.

Phonetic Systems’ technology incorporates a highly accurate, conversational-style search method that ensures callers receive correct phone numbers and that calls are routed to proper destinations. The conversational search acts like a live operator. If it is unsure of what a caller said, it dynamically selects questions based upon the context of that particular caller’s input. Features like natural language techniques, spell by name, selection between listings with identical names, progressive help (such as asking a caller to speak louder or repeat specific pieces of information), are just some of the robust error recovery capabilities. In addition, the technology has the ability to search and differentiate between similar sounding listings and the use of nicknames.

Nuance has also announced enhancements to its line of speech recognition and verification software. Nuance Enterprise Solutions comprise a family of suites designed to reduce the cost and the time required to build and deploy V-Commerce, or voice-enabled e-commerce, interfaces to leading packaged applications.

Other companies like Webley and Wildfire are using speech technology to create personal assistants that use speech recognition to help users manage fax, phone, and e-mail communications. Efforts like these should help broaden the speech marketplace, resulting in more developers using speech technology and hence the deployment of more speech applications.
— Carol Drzewianowski, CTI Magazine

Not “Just” Browsing

Browser-based technology has become ubiquitous, showing up in applications that have nothing to do with surfing the Web. It makes sense. Browsers are familiar to a large — and growing — segment of the population, so workers using browser-based applications need less training to learn the software. And browser-based technology is powerful and flexible in sorting and distributing information, as well as making data and resources widely available, often without regard to location.

Lately, we’ve seen a lot of browser-based contact center applications and strategies — and with good cause. In essence, contact centers work more productively by giving managers and agents more widespread access to call center resources and data and greater ability to monitor and manage performance and scheduling. Those two areas — performance tracking and scheduling—have driven many of the solutions that we are seeing recently. Often, the distinction between solutions lies in whether the solutions aim to give managers greater monitoring and control capabilities or to enable agents to self-manage their time and information more effectively. The most effective solutions seem to be those that combine aspects of both.

ISC’s Irene is one such solution. Irene uses scalable browser-based forecasting and scheduling software for call centers, allowing managers and authorized agents to view and print schedules, request days off, and perform other administrative tasks. Because it uses an Internet-based architecture, Irene can be used for multiple call centers and multiple locations without excessive software installation costs. Added functionality comes from its use of industry standards like SQL, XML, HTML, and ODBC, and from the fact that data can be exported to other desktop applications including Microsoft Word and Excel.

Siemens also has introduced a call center management solution that leverages Internet-based technology to deliver expanded access and management capabilities. Siemens Vantage+3.0 is an Internet-accessible, real-time agent tracking tool that allows mobile call center managers to view real-time, near-real-time, and historical reports on such critical areas as service level, calls received, calls answered and abandoned, average speed to answer, and answer percentage.

Geared toward remote call center managers and companies that outsource call center activities who need the flexibility to monitor and track call center performance via the Internet, Siemens’ solution adds functionality by bundling into the package viewing and reporting capabilities that allow managers to effectively organize the information they access. Reports can be assembled in Vantage+’s color-coded floor plan view, which enables managers to obtain an agent’s call statistics by clicking on their seat position within the call center, or generated onscreen in row/column format. The reports can be imported to any ODBC-compliant applications so the information can be integrated with other business applications.

Vantage+3.0 runs on one NT SQL server, reducing hardware requirements, and is user configurable, meaning that call center managers can change floor plans or add agents. Call diverting allows managers to drag and drop calls from one queued group to another, or to a specific call center agent. Call qualifying, another handy function, lets agents code incoming calls so they can later be monitored by type of call.

IEX also has incorporated browser technology into its workforce management solution. IEX announced the addition of the Agent WebStation to its TotalView Workforce Management software, a client/server system that runs on Windows, using an ODBC reporting interface. TotalView integrates forecasting and scheduling, attendance and agent-productivity tracking, payroll input, vacation and holiday planning, Web-enabled schedule and statistics viewing, meeting scheduling, and adherence viewing. For multiple site operations, TotalView interfaces to IEX’s TotalNet Call Routing as part of a virtual call center.

The Agent WebStation module allows call center agents to utilize a Web browser to view call statistics and to access their schedules from a desktop computer. To access the Agent Webstation, call center agents connect to the TotalView server and are prompted for an ID and password. An IEX-developed Java applet maintains data security.

Once an agent is logged on, the Agent WebStation provides up-to-date information to the agent, tracking call handling efficiency and automating the distribution of work schedules and schedule changes through its StatsViewer and ScheduleViewer modules. StatsViewer displays the agents’ personal call statistics, the statistics of a group of agents managed by the same supervisor, and statistics for groups that are scheduled and managed together. Agents may personalize their own StatsViewer modules to view data fields such as number of calls, talk time, work time, average handle time, average talk time, etc. Supervisors can also define the fields that will be available for display to agents.

ScheduleViewer allows agents to access and manipulate their work schedules, alerting them to new schedules or changes and allowing them to confirm that they have seen schedules. Automating simple, repetitive processes of that nature makes better use of both agents’ and supervisors’ time.
— Grace Pisano,CTI Magazine