Tellme Architecture

Tellme provides speech processing for voice applications. Shown here is the architecture of a system in which Tellme provides outsourced speech processing for a voice application operated by a customer, such as a credit card processing firm or airline.

A call comes to Tellme via a phone system, generally one operated by a Tellme partner (currently AT&T, Global Crossing, or Verizon). Tellme generates voice prompts and recognizes the caller’s speech and touch-tone inputs using speech engines and Voice XML pages retrieved from the customer’s voice application over a secure Web (HTTPS) connection. Voice XML is a markup language similar to HTML that specifies prompts, expected responses, and actions to take on recognizing a response. Like HTML pages, Voice XML pages can be static or can be generated dynamically by the voice application, using data from a customer relationship management (CRM) system or other internal applications. Tellme uses speech engines from several vendors, currently IBM, Microsoft, and Nuance.