Ambisonics Direction-of-Arrival Estimation Using Angle Segments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dual-lens image capture devices often record insufficient audio for rear-facing lenses, particularly when users narrate scenes, necessitating improved methods for determining the direction of audio arrival.
Innovation Solution
Implementing Ambisonics channels with multiple microphones and a processor to divide time intervals into blocks, estimate azimuth and elevation angles based on the highest number of angle estimates in defined segments, and determine the direction of arrival using these angles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single audio stream is recorded for dual-lens capture devices, then the audio is sufficient for the forward-facing lens, but it is insufficient for the rear-facing lens when the user is narrating the scene
Solution Approach 1:
The patent segments the audio processing into multiple independent channels corresponding to different spatial directions (front, back, left, right). By dividing the audio field into distinct segments and processing each separately, the system can selectively enhance audio from specific directions, making the audio sufficient for both forward-facing and rear-facing lenses simultaneously.
2Measurement precision
If Ambisonics channels are processed to determine direction of arrival, then audio directionality is improved, but computational complexity increases due to block processing and angle estimation
Solution Approach 1:
The time interval is divided into multiple blocks, and the azimuth and elevation spaces are segmented into discrete segments. This segmentation allows the system to process smaller, manageable units independently and combine results, improving measurement precision while controlling computational complexity through systematic breakdown of the processing task.
Solution Approach 2:
The patent processes a defined number of blocks within a time interval and uses majority voting among angle estimates. By processing partial sets of data and using statistical aggregation, the system achieves high directional accuracy without requiring exhaustive processing of all possible data, thus balancing precision with computational efficiency.
Data Source
AI summary
Devices and methods for determining a direction of audio arrival from Ambisonics channels using azimuth and elevation segments is described herein. A method includes generating multiple blocks of samples from Ambisonics signals for a time interval, determining an azimuth angle estimate and an elevation angle estimate for the time interval when a defined number of blocks in the multiple blocks of samples are valid, generating the azimuth angle estimate based on maximum number of azimuth angle estimates present in an azimuth segment amongst a defined number of azimuth segments, and generating the elevation angle estimates based on maximum number of elevation angle estimates present in an elevation segment amongst a defined number of elevation segments, where the direction of arrival of the Ambisonics signals is based on the azimuth angle estimate and the elevation angle estimate.