Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Method and System for Automatically Adapting Voice Portal to the Caller's Voice Characteristics

IP.com Disclosure Number: IPCOM000028407D
Original Publication Date: 2004-May-14
Included in the Prior Art Database: 2004-May-14
Document File: 2 page(s) / 11K

Publishing Venue

IBM

Abstract

The disclosed is a method and system to automatically adapt voice portal to a caller's voice characteristics. The specific aspects of the voice portal application which need adaptation are acoustic model, audio files and/or text-to-speech, and grammars. The detection of a caller's voice characteristics is done by the voice browser under the cover though the existing voice characteristic related technology.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 2

Method and System for Automatically Adapting Voice Portal to the Caller's Voice Characteristics

Voice portal has become an important means for people to receive information anytime and anywhere. However, the voice portal as an extension from the visual portal, currently supports the personalization only through the content organization. Although all the people in United States are reading the same English in the US locale, they are speaking English differently, for example, with a Northern accent, Southern accent, Asian accent, Indian Accent and so on. Different accent groups may have different ways to express ideas. However this information is not easily solicited from the user. A voice portal that can automatically adapt to a caller's voice characteristics is very desirable. Also the voice portal serves as a gateway for the caller to reach almost any voice application on the internet. There is another problem to make all the voice applications aware of the caller's voice characteristics and to adapt correspondingly if supported.

The disclosed is a method and system to automatically adapt voice portal to a caller's voice characteristics. The specific aspects of the voice portal application which need adaptation are acoustic model, audio files and/or text-to-speech, and grammars. The detection of a caller's voice characteristics is done by the voice browser under the cover though the existing voice characteristic related technology. Based on the characteristics, the browser will switch to corresponding acoustic model and the Text-to-Speech engine. The browser will submit the characteristics information to every voice application the caller wants to access through a hidden http request parameter. The voice application could be on any server on the internet. The voice application can provide the adaption on the audio files and grammars based on the characteristics, if it is supported. Otherwise, it will serve the default ones as normal.

The advantage of this concept is that it provides a truly personalized voice portal service to the users. By adapting the voice portal application towards a caller's voice characteristics, it increases the ease of use of the applications. The voice characteristics information can be used to profile the callers to provide statistics data and/or improve the application thereafter. Also the voice characteristics information can be carried in to Computer Telephony Integration world to match an agent with similar voice characteristics upon transfer.

The disclosed system consists of the following parts: a set of existing feature vectors for corresponding to different voice characteristics,...