Distributed Microphone Array Voice Awakening Sub-Array Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In intelligent household scenarios, multiple devices often respond simultaneously when awakened by a user's voice, leading to poor user experience due to the lack of effective methods to determine the intended device for interaction.
Innovation Solution
A method and apparatus for processing voice signals using a distributed microphone array, which involves obtaining awakening voice signals, determining frequency domain signals, and calculating cross-correlation functions to identify the awakened sub-array most suitable for interaction, thereby improving the accuracy of device awakening decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple intelligent devices in the household respond simultaneously to a user's voice, then the coverage of voice recognition is improved, but the user experience deteriorates due to lack of device selection
Solution Approach 1:
The distributed microphone array is divided into multiple sub-arrays, each corresponding to a specific intelligent device. This segmentation allows the system to identify which sub-array (and thus which device) the user intends to interact with, resolving the ambiguity of simultaneous responses while maintaining broad coverage.
Solution Approach 2:
The patent replaces mechanical/device-based awakening decisions with acoustic field analysis. By using cross-correlation functions to analyze the acoustic characteristics of awakening voices across different sub-arrays, the system determines the most likely target device without requiring manual selection or complex device communication protocols.
2Device complexity
If traditional time domain cross-correlation methods are used for awakening voice detection, then the implementation is simple, but the detection precision deteriorates due to interference, noise, and reverberation
Solution Approach 1:
The patent substitutes time-domain cross-correlation with frequency-domain cross-correlation. This transformation allows the system to analyze the spectral characteristics of awakening voices, making the detection more robust against interference, noise, and reverberation while maintaining reasonable computational complexity through efficient frequency-domain algorithms.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy of device awakening decisions by identifying the correct sub-array based on cross-correlation functions, improving reliability and reducing interference, noise, and reverberation resistance, ensuring the intended device is selected for user interaction.
Implementation Method 1
determining, for each sub-array, a frequency domain signal corresponding to each awakening voice signal of the sub-array
Implementation Method 2
a first cross-correlation function between every two frequency domain signals; and determining an awakened sub-array to be awakened by a corresponding awakening voice signal in the sub-arrays based on each first cross-correlation function for each sub-array
Data Source
AI summary
A method and an apparatus for processing voice are provided. The method is applied to a decision-making device in communication with a distributed microphone array and the distributed microphone array comprises a plurality of sub-arrays. The method comprises: obtaining, for each sub-array, an awakening voice signal received by each microphone of the sub-array; determining, for each sub-array, a frequency domain signal corresponding to each awakening voice signal of the sub-array, and a first cross-correlation function between every two frequency domain signals; determining an awakened sub-array based on each first cross-correlation function for each sub-array.