Voice Input Device with Fixed Microphone Array for Directional Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice input devices face challenges in maintaining noise suppression performance and accurate voice recognition after movement, as the positional relationship between the microphone array and the noise direction changes, requiring re-learning of noise information, which can lead to deteriorated performance and failed recognition.
Innovation Solution
A voice input device with a fixed microphone array and a movable part that estimates the utterance direction using a signal processing unit, allowing the device to adjust its orientation based on the estimated direction without altering the microphone array's position, thereby maintaining effective noise suppression and recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the device turns to the utterance direction, then the device can face the speaker for better interaction, but the positional relationship between the microphone array and noise direction changes, breaking the learned noise information
Solution Approach 1:
The device is divided into a fixed part (housing with microphone array) and a movable part (display unit). The microphone array remains fixed in the housing to maintain its positional relationship with noise sources, while the display unit can independently rotate to face the speaker. This segmentation allows the noise suppression system to maintain reliability while the device can still orient toward the speaker for better interaction.
2Productivity
If the device turns to the utterance direction, then the device can engage the speaker directly, but sufficient learning time for noise information cannot be secured, causing noise suppression performance to deteriorate
Solution Approach 1:
The noise information is learned and stored in advance in the correlation matrix before the device needs to respond. The system continuously updates the correlation matrix with noise information from the environment, so that when the device turns to face the speaker, the noise suppression algorithms already have pre-learned noise characteristics to work with, eliminating the need for real-time learning during the turning action.
Data Source
AI summary
The present technology relates to a voice input device and method that facilitate estimation of an utterance direction. The voice input device includes a fixed part disposed at a predetermined position, a movable part movable with respect to the fixed part, a microphone array attached to the fixed part, an utterance direction estimation unit that estimates an utterance direction on the basis of a voice from an utterer that is input from the microphone array, and a driving unit that drive the movable part according to the estimated utterance direction. The voice input device can be used by installation in, for example, a smart speaker, a voice agent, a robot, and the like.