Sound Recognition Using Microphone Attitude for Position-Independent Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sound recognition methods struggle to accurately identify similar or identical sounds and estimate sound sources when the position of the microphone changes, as they rely on the positional relationship between the sound source and microphone, which is not reliable in such scenarios.
Innovation Solution
A sound recognition apparatus and method utilizing at least two microphones, an attitude detection unit, a self-position estimation unit, and a sound source estimation unit to create a sound environment map, which includes estimating microphone positions and sound source directions based on detected attitudes and signals, allowing for accurate identification and localization of sound sources even when the microphone's position changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sound recognition is performed based on the positional relationship between the sound source and the microphone, then the number of sound sources, spatial existence range, existence period, and direction of arrival can be specified, but the position or direction of the sound source cannot be recognized when the position of the measurement microphone is changed
Solution Approach 1:
The patent introduces a new dimension of reference by using the relative positional relationships between multiple microphones as an additional reference frame. Instead of relying solely on the absolute position between sound source and single microphone, the system uses the spatial configuration of multiple microphones to create a stable reference that remains valid even when the overall microphone array position changes.
Solution Approach 2:
The system makes the microphone array itself serve multiple functions: both as the detection element for sound signals and as a reference frame for position estimation. The relative positions between microphones serve as a universal reference that works regardless of the array's absolute position in space.
2Measurement precision
If characteristics of sound itself are extracted for identification, then sound features can be extracted, but similar sounds or a plurality of identical sounds cannot be identified from each other
Solution Approach 1:
The patent segments the sound recognition problem into two independent components: sound feature extraction (what the sound is) and position estimation (where the sound is from). By separating these functions, the system can extract general sound characteristics while simultaneously capturing spatial information that enables discrimination between similar sounds from different locations.
Solution Approach 2:
The system introduces position information as an intermediary element that bridges the gap between sound feature extraction and sound source identification. Even when sound features are similar, the positional information derived from multiple microphone signals serves as an additional discriminator to distinguish between different sound sources.
Data Source
AI summary
A sound recognition apparatus (100) includes at least two microphones (1) that detect a surrounding sound, an attitude detection unit (4) that detects attitudes of the microphones (1), a self-position estimation unit (52) that estimates positions of the microphones (1) based on the attitudes of the microphones (1) and a signal of the sound detected by the microphone (1), a sound source estimation unit (53) that estimates a direction of the sound source (200) of the sound based on the attitudes of the microphones (1) and the signal of the sound detected by the microphone (1), and a sound environment mapping unit (54) that creates a sound environment map based on the positions of the microphones (1) and the direction of the sound source (200), the sound environment map displaying at least a position of the sound source (200).