Sound Collection Beam Signal Selection for Speaker Orientation Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional audio conference systems with integrated microphones and loudspeakers face challenges in accurately detecting speaker orientation due to diffraction sounds, leading to poor sound collection and output quality.
Innovation Solution
A sound emission and collection apparatus with multiple microphones arranged in specific patterns, employing delay and amplitude processing to generate sound collection beam signals with varying directivity, and selecting the signal with an energy ratio exceeding a predetermined threshold to isolate the speaker's orientation, thereby minimizing diffraction interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a microphone is placed close to a loudspeaker in an integrated apparatus, then the apparatus structure is simplified and compact, but diffraction sound from the loudspeaker is largely included in the sound collection signal, reducing speaker orientation detection accuracy
Solution Approach 1:
The patent divides the sound collection function into multiple independent microphones arranged in specific patterns, allowing differential processing of signals from different microphones to eliminate diffraction sound while preserving speaker orientation information
Solution Approach 2:
The patent applies different processing methods to different microphone signals based on their spatial positions and orientations, using delay and amplitude processing to create beam signals with specific directivity characteristics that are optimized for detecting speakers from particular directions
2Object-generated harmful factors
If diffraction sound volume is large and utterance sound volume is small, then the signal from the speaker orientation can be detected, but the speaker orientation cannot be accurately detected to accurately collect the sound from the orientation
Solution Approach 1:
The patent converts the harmful diffraction sound into a useful reference by using the fact that diffraction sound appears in all microphone signals with similar characteristics, allowing the system to subtract or eliminate this common component through differential processing to reveal the true speaker orientation signal
Solution Approach 2:
The patent introduces beam signal processing as an intermediary step between raw microphone signals and final speaker orientation detection, using delay and amplitude processing to create intermediate signals that enhance speaker orientation information while suppressing diffraction sound
Data Source
AI summary
A level ratio calculation circuit calculates average signal level data of signal level data corresponding to each sound collection beam signal, and calculates a level ratio between the average signal level data and each of the signal level data. Since a diffraction sound is substantially equal to all the signal level data, a diffraction sound component of the average signal level data also becomes substantially equal. On the other hand, a collection sound from a speaker is specific to the signal level data of the corresponding sound collection beam signal. Therefore, at the level ratio, the portion corresponding to the diffraction sound is flat and a data level becomes high locally in only the portion corresponding to the collection sound. By using this, the sound collection beam signal including the collection sound is detected.