Microphone Array Direction Setting for Speech Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing speech recognition systems with multi-directional microphone arrays do not activate direction settings prior to receiving a wakeup command, leading to unintended or missed activations due to interference during operation in a directionless setting.

Innovation Solution

A sensor detects the relative location of a user within one of multiple voice pickup areas of the microphone array, triggering the activation of a direction setting to enhance speech recognition, which includes directing a beamforming function towards the user's location to improve the detection of wakeup commands.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the multi-directional microphone array operates in a directionless setting, then the system can detect wakeup commands from any direction, but the system experiences interference and unintended activations

Engineering Contradiction:
Improvevoice pickup area coverageVSAvoidactivation accuracy
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The system performs preliminary action by activating the direction setting in advance based on sensor detection of user presence and location, before the wakeup command is actually received. This prevents interference during operation by establishing the appropriate directional configuration beforehand.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses feedback from sensors that detect user location and voice pickup area information to dynamically adjust the microphone array's directional setting. This feedback loop allows the system to adapt its operation mode based on real-time environmental conditions, resolving the contradiction between omnidirectional coverage and activation accuracy.

Inventive Principle:
Principle #23Feedback

2Device complexity

If the direction setting is activated after receiving the wakeup command, then the system structure remains simple, but unintended or missed activations occur due to interference

Engineering Contradiction:
Improvecontrol mechanism simplicityVSAvoidwakeup command detection accuracy
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The direction setting is activated in advance before the wakeup command is received, based on sensor detection of user presence. This preliminary activation eliminates interference that would otherwise cause unintended or missed activations, while the control mechanism remains relatively simple by using automatic sensor-triggered activation.

Inventive Principle:
Principle #10Preliminary action

3Ease of operation

If the multi-directional microphone array uses a directionless setting, then the system is easy to operate, but speech recognition is not optimized for specific directions

Engineering Contradiction:
Improvesystem operation simplicityVSAvoidspeech recognition accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The system dynamically adjusts its operation mode based on sensor detection. When a user is detected in a specific voice pickup area, the system automatically activates the corresponding direction setting to optimize speech recognition accuracy for that direction. This dynamic adaptation maintains ease of operation while improving measurement precision when needed.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system performs self-service by automatically detecting user location through sensors and activating the appropriate direction setting without requiring manual intervention. This autonomous operation maintains simplicity for the user while optimizing speech recognition accuracy based on the detected spatial conditions.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3889956B1Systems and methods for automatic speech recognition
Publication Date: 2024.04.24 ADEMCO INC
  • EP3889956B1 patent drawingFigure 1~2
  • EP3889956B1 patent drawingFigure 3~4
  • EP3889956B1 patent drawingFigure 5

AI summary

Systems and methods for automatic speech recognition are provided. Some methods can include a sensor detecting a relative location of a user within one of a plurality of voice pickup areas of a multi-directional microphone array and the multi-directional microphone array activating a direction setting of the multi-directional microphone array based on the relative location of the user within the one of the plurality of voice pickup areas.