Speech processing apparatus and method

摘要

A speech processing apparatus and method. The speech processing apparatus includes a microphone to receive a speech signal, an analog/digital converter to convert the speech signal generated by the microphone into a digital speech signal, and an automatic gain controller to calculate an average value of the magnitude of the digital speech signal generated by the analog/digital converter in a plurality of frames, to determine in which region of a speech signal band the average value is located, the speech signal band being divided into a plurality of regions according to the strength of speech, and to adjust gain according to a location of the average value on the speech signal band so that the strength of speech has a level of an optimal region capable of processing the speech signal. Accordingly, speech recognition may be maximized without being constrained by the distance of a speech source.

主权项

1. A speech processing apparatus, comprising:
a microphone configured to receive a speech signal; an analog/digital converter configured to convert the speech signal generated by the microphone into a digital speech signal; and an automatic gain controller configured to calculate an average value of the magnitude of the digital speech signal generated by the analog/digital converter in a plurality of frames, where the number of frames is predetermined, determine in which region of a speech signal band the average value is located, the speech signal band being divided into a plurality of regions according to a strength of speech based on a dynamic range of the microphone, and adjust gain according to a corresponding one of the plurality of regions to which a location of the average value belongs so that the strength of speech has a level of an optimal region where the speech signal is capable of being processed, wherein the speech signal band is divided into a plurality of regions based on a minimum limit capable of being detected by the microphone, a maximum limit level capable of being detected by the microphone, low and high levels of the optimal region, and a median point of the speech signal band.