Amazon Technologies, Inc.

Xiaoxue Li

Philip Ryan Hilmes

Trausti Thor Kristjansson

Amit Singh Chhetri

Rui Wang

Systems and methods for microphone degradation detection and compensation are disclosed. For example, microphones of an electronic device may capture audio and generate corresponding audio data, such as during a period of time where only ambient noise is present. Sound intensity level value differences between audio data from the various microphones may be determined and when one or more of the sound intensity level value differences satisfies a threshold amount, the microphone associated with the variant sound intensity level value may be determined to be degraded. The sound intensity level value difference may be compensated for, such as by utilizing sound boosting techniques and/or modifying parameters of a beamforming component.

Microphone degradation detection and compensation

An audio capture device that incorporates a beamformer and beam-specific trigger word detection. Audio data from each beam is processed by a low power trigger word detector, such as a neural network or other trained model to detect if audio data (such as an audio frame or feature vector corresponding thereto) likely includes part of a trigger word. The beam that either most strongly represents a trigger word portion or represents a trigger word portion most early in time may be selected for further processing such as speech processing or confirmation by a more robust power intensive trigger word detector.

Trigger word based beam selection

Features are disclosed for estimating reverberation an audio signal without prior knowledge of the sound captured in the audio signal and without prior knowledge of the environment in which the sound was captured. This reverberation estimation may be referred to as 'blind estimation.' The reverberation may be blindly estimated using autocorrelation of the sound intensity of the signal with respect to time. The blind estimation can be used to decide whether or not to mitigate the effects of reverberation in the signal prior to performing subsequent processes, such as automatic speech recognition.

Growing community of inventors

Mountain View, CA, United States of America

Xiaoxue Li