Speech recognition system for automatically controlling input level and speech recognition method using the same

Abstract

A speech recognition system, comprising: a speech pick-up element, used to pick up the speech from an external speaker; a speech level controller, used to receive the picked-up speech with the speech input level provided by the speech recognizer, and the received speech The speech is output to the speech recognizer; the speech detector is used to detect the period of the speech signal required for speech recognition from the speech output from the speech receiver; the speech saturation detector is used to determine the period of the detected speech signal based on the threshold value Whether the voice signal is saturated; input level determiner, for when the voice signal in the voice signal period is saturated, determine the new voice input level, and output the information about the new voice input level to the voice receiver, thus the voice receiver Speech is received in an unsaturated state. A voice recognition method includes steps corresponding to the functions of the above system.

Description

technical field [0001] The present invention relates to a speech recognition system, and more particularly, to a speech recognition system capable of controlling the input level of a speech according to whether a speech signal period of an input speech is detected and whether the speech signal in the speech signal period is saturated, and a speech recognition system method. Background technique [0002] In general, a speech recognition system or method generates a feature vector of an input speech by various analysis methods using a frequency analysis scheme, and recognizes the speech using the generated feature vector. Such a speech recognition system or method uses one of several speech recognition schemes that use the energy of the input speech signal. [0003] In such a speech recognition system or method using energy of an input speech signal, in order to recognize speech, the energy of the input speech signal is normalized to minimize deviations therein. In this rega...

AI Technical Summary

Problems solved by technology

[0004] In the existing speech recognition system or method, there is such a problem: when the speech detection fails due to the input level of the speech signal is too low, or when the speech input level deviates within a certain period of time due to the speech input level is too high Speech recognition rate may drop when voice input resolution

However, in the above speech recognition system or method, each individual user must manually control the speech input level

Images