Contact:
Newsfox Desk
Phone: + 43 - 1 - 811 40 - 319
E-Mail: editor@newsfox.com
| Pressbox | 
|
Berlin (pte032/29.04.2003/14:46) - Semiconductor producer Intel http://www.intel.com introduced a programme at the Developer Forum in Berlin, which they claim can teach computers to read lips.
Designed by the Intel research department, the system is based on a face recognition algorithm, as well as audio-visual speech recognition (ASVR). The software package has been published as part of an open source license for developers who want to improve the language capabilities of their computers.
One of the biggest obstacles to computer speech recognition has been the presence of background noise. The new system instead combines ASVR and face recognition to follow the speakers lip movements. According to Intel, the synchronisation of video data with language recognition permits more precise speech recognition.
The ASVR software is part of the Intel OpenCV Computer Vision Library. The toolbox consists of over 500 image processing functions, with which researchers can develop image recognition applications. The aim of this new technology, says Intel, is to promote a more natural form of interaction between users and computers.
"Recognition among humans seldom depends on a single type of information. We make decisions based on the combination of information from a number of sources," said Justin Rattner, Intel Senior Fellow of the Enterprise Platform Group and director of Intel Microprocessor Research Labs.
Rattner believes the audio-visual recognition code now taken up in Intels OpenCV Library will promote research and development in image-supported speech
Further information on ASVR and the free licensing agreement can be found at www.intel.com/research/mrl/research/avscr.htm .
(end)
|
|
