keyword spotting


Also found in: Wikipedia.

keyword spotting

[′kē‚wərd ‚spät·iŋ]
(engineering acoustics)
An approach to task-oriented speech understanding through detecting a limited number of keywords that would most likely express the intent of a speaker, rather than attempting to recognize every word in an utterance.
References in periodicals archive ?
It's vital you keep your market knowledge up to date segmented by Applications [Application I, Application II, Application III & ], Product Types [Isolated Word Recognition, Keyword Spotting & Continuous Speech Recognition] and major players.
functions such as image recognition or keyword spotting without sending
Keyword spotting technology makes a substantial part of such systems.
As a result, we were able to build an autonomous version of the game by extending classic, example-based keyword spotting to handle overlapping speech and historical context.
CSpotter is a voice recognition engine that supports low-power always listening with keyword spotting.
Echo uses on-device keyword spotting to detect the wake word.
Fortemedia has completed production of its iM401D low power, keyword spotting voice trigger and control IC.
With an extremely noise robust and accurate solution that responds quickly and at ultra-low power consumption, the TrulyHandsfree[TM] trigger technology has become the most widely adopted keyword spotting technology in the speech industry.
"We're not doing keyword spotting, because we're looking at every word said, including the acoustics of how it was said, the underlying sentiment," he continues.
Joe was sure to point out how the company's offering features recording, UC; real-time speech analytics; quality management; multichannel communications including voice, web chat and social media; routing; and real-time speech analytics with keyword spotting. There is also integration with Salesforce and other services our of the box.
The company says traditional keyword spotting solutions fail to truly understand the rich meaning and conceptual patterns within user-generated content.