Lip reading
A new software, developed by computer scientists at Carnegie Mellon University (CMU) at Pittsburgh, usa, will aid computers to recognise speech through lip reading. "The new software, NLips, will reduce the error rate of speech recognition software in noisy environments,' says Alex Waibel of the CMU. Normal speech recognition software is 60 per cent successful when the surroundings are noisy, while the new software is 85 per cent successful. NLips, like most speech recognition systems, breaks down speech into sound chunks called phonemes. But significantly, the new software combines information from lip movements and uses computer-mounted cameras to record lip sequences ( New Scientist , Vol 163, No 2199).
Related Content
- First food: business of taste
- Dietary guidelines for India 2024
- Draft Offshore Areas Mineral Trust Rules, 2024
- Water (Prevention and Control of Pollution) Amendment Act 2024
- Meghalaya Waterbodies (Conservation and Protection) Amendment Guidelines, 2023
- Notification on whole UT of Ladakh as Air Pollution Control Area in exercise of powers conferred by Section 19(1) read with Section 6 of Air Act, 1981