Hearing System Own-Voice Detection Using Spectral SPL Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting own voice content in hearing devices require specialized components and complex processing, leading to inefficient and resource-intensive operations.
Innovation Solution
A hearing system utilizing ipsilateral and contralateral microphones to determine sound pressure levels and symmetry levels of spectral portions of audio signals, combined with machine learning algorithms, to accurately identify own voice content without additional hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods use specialized components and complex processing to detect own voice content, then detection accuracy is improved, but device complexity and computational requirements increase
Solution Approach 1:
The patent creates a virtual copy of the user's voice by generating synthetic own-voice signals from recorded speech samples. This synthetic voice model is then used for comparison with incoming audio signals to detect own-voice content, eliminating the need for complex specialized components while maintaining detection accuracy
Solution Approach 2:
The patent introduces a synthetic own-voice signal as an intermediary between the user's actual voice and the detection system. This synthetic signal serves as a reference model that mediates the comparison process, simplifying the detection mechanism while improving accuracy through dynamic adaptation
2Measurement precision
If conventional methods use specialized components and complex processing to detect own voice content, then detection accuracy is improved, but energy consumption increases
Solution Approach 1:
By creating a synthetic copy of the user's voice rather than using complex real-time analysis, the system reduces computational burden and energy consumption while maintaining detection accuracy
Solution Approach 2:
The system performs preliminary recording and synthesis of the user's voice characteristics during quiet periods, preparing the synthetic own-voice model in advance. This preliminary action reduces the need for complex real-time processing during active speech detection, thereby lowering energy consumption
3Productivity
If conventional methods detect own voice content, then audio processing can be optimized, but naturalness of user's own voice may be compromised
Solution Approach 1:
The patent applies different processing strategies to different audio sources by identifying own-voice content and applying specific enhancement techniques only to those signals. This local differentiation allows optimized processing for own voice while preserving the natural characteristics of other audio sources
Solution Approach 2:
The synthetic own-voice model enables the system to recognize and selectively process the user's voice differently from other sounds, allowing for natural sound reproduction while optimizing overall audio processing efficiency
Data Source
AI summary
An exemplary hearing device is configured to determine a first sound pressure level (SPL) of a first spectral portion of an ipsilateral audio signal representative of audio content and detected by an ipsilateral microphone associated with an ipsilateral ear of a user, the first spectral portion having frequencies included in a first frequency range. The device may further determine a second SPL of a second spectral portion of the ipsilateral audio signal, the second spectral portion having frequencies included in a second frequency range that is higher than the first frequency range. The device may further determine, based on a determination that the first SPL is greater than the second SPL by at least a threshold SPL amount, that the audio content comprises own voice content representative of a voice of the user.