Acoustic echo cancellation processing based on feedback from speech recognizer

摘要

An automatic speech recognition engine receives an acoustic-echo processed signal from an acoustic-echo processing (AEP) module, where said echo processed signal contains mainly the speech from the near-end talker. The automatic speech recognition engine analyzes the content of the acoustic-echo processed signal to determine whether words or keywords are present. Based upon the results of this analysis, the automatic speech recognition engine produces a value reflecting the likelihood that some words or keywords are detected. Said value is provided to the AEP module. Based upon the value, the AEP module determines if there is double talk and processes the incoming signals accordingly to enhance its performance.

主权项

1. A method of processing signals within an apparatus comprising an acoustic-echo processing module, an automatic speech recognition engine, a loudspeaker and at least one microphone, the method comprising:
receiving, at the automatic speech recognition engine from the acoustic-echo processing module, an acoustic-echo processed signal, wherein the acoustic-echo processed signal is based at least in part upon processing a first signal received at the microphone that comprises a component from the loudspeaker, the processing to cancel at least a portion of the component from the first signal; determining, by the automatic speech recognition engine, audio content within the acoustic-echo processed signal; based at least in part upon the audio content determined to be within the acoustic-echo processed signal, producing, by the automatic speech recognition engine, a value indicative of a confidence that one or more spoken commands are present in the audio content; determining that the value exceeds a threshold value; providing the value to the acoustic-echo processing module; adjusting one or more control parameters of the acoustic-echo processing module based on the value; and processing a subsequent signal received by the acoustic-echo processing module, the processing comprising canceling at least a portion of a component from the loudspeaker corresponding to the subsequent signal.