Unisys has introduced its Natural Language Speech Assistant application suite to SA`s telecommunications industry, providing the market with technology that will improve the country`s interactive voice response (IVR) application development environment.
The Unisys suite combines the company`s existing natural language processing technology with third-party speech recognition applications to create a set of tools that can be used to enhance the use of IVR.
"Natural speech recognition technology has the power to enhance user interaction by removing the impersonal `press one for account enquiries`, and giving users the freedom to voice their requirements in their language of choice," says Dr Bill Scholz, the architect director of voice and business mobilisation solutions engineering at Unisys Global Communications.
The Prompt Preparation tool in the suite enables the creation of recording scripts to assist in presentation, review and integration of professional talent into the speech application. The Design Review tool facilitates the development of proposal documents showing various scenarios the application will present to a caller, and dialogue design documents summarising the application and its architecture.
Call Flow Verification and Usability tools allow "use scenarios" to be produced as HTML-formatted reports; a Wizard of Oz simulator enables usability studies; and the Application Deployment tool allows the production of sound files and grammars to constrain speech recognisers, the representation of application call flows in XML, and the implementation of Java code for the application`s structure.
"Building speech applications can be an unnecessarily complicated and expensive undertaking if the right tools are not available," says Dean Baker, acting business unit head: Telecommunications at Unisys Africa. "This latest Unisys suite un-complicates the process, making speech application development, and the introduction of natural language processing into the IVR environment, more accessible to the South African market. It also enables speech application developers to more easily create solutions that encompass the countries diverse language requirements."
The suite permits the representation of a speech application as a collection of dialogue states where each state consists of a simple conversational interchange between the computer and the user. Multiple dialogue states are interconnected by response-contingent links, forming a network of states that represent the call flow.
It provides facilities for managing complex variables and sound files, and a mechanism for simulating the application prior to the generation of actual code. An easy-to-understand spreadsheet metaphor is used to express the relationships between words and phrases that define precisely what the end-user is expected to say, and what those utterances mean to the application.
"The Dialogue Server renders the abstract user interface published by the developer into VoiceXML, ensuring the user interface is presented to the user in a rehearsed, predictable and stable manner," Dr Scholz concludes. "This server manages the low-level protocols, interface rendering, protocol `dialects`, variable resolution, and deployment structure."
Share
Unisys is a worldwide information technology services and solutions company. Our people combine expertise in consulting, systems integration, outsourcing, infrastructure and server technology with precision thinking and relentless execution to help clients, in more than 100 countries, quickly and efficiently achieve competitive advantage. For more information, visit www.unisys.co.za.
Editorial contacts