Binaural Beamformer Using Interaural Differences for Directional Audio
Solution Overview
Problem
Existing binaural assistive listening devices face challenges in producing directional sound outputs effectively, particularly in complex environments with low signal-to-noise ratios, due to limitations in broadside array configurations, adaptive Least Minimum Square (LMS) systems, and Blind Source Separation (BSS) schemes, which struggle with frequency dependence, computational overload, and noise rejection.
Innovation Solution
A novel binaural beamformer approach that utilizes interaural time and level differences to produce directional responses, employing filter weights based on signal similarity calculations and head-related transfer functions to cancel low and high frequency sounds, offering improved noise rejection and reduced computational demands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If broadside array configurations are used, then directional response is produced for low-frequency sounds, but the system becomes ineffective for high-frequency components
Solution Approach 1:
The patent divides the frequency range into low-frequency and high-frequency components, applying different processing strategies to each. For low frequencies, broadside array configurations are used to exploit interaural time differences, while for high frequencies, the system exploits interaural level differences and head shadowing effects. This segmentation allows the system to maintain effectiveness across the entire audible spectrum.
Solution Approach 2:
The patent applies different beamforming strategies to different frequency bands. The low-frequency processing uses one set of parameters and characteristics, while high-frequency processing uses different parameters adapted to the physical characteristics of the head and microphone spacing at higher frequencies. This local optimization ensures maximum directional response at each frequency range.
2Adaptability or versatility
If LMS systems with VAD are used, then directionality is produced independently of frequency, but computational complexity and dependence on VAD performance increase
Solution Approach 1:
The patent extracts and exploits the natural spatial acoustics of the head itself as the directional filter, rather than relying on complex adaptive algorithms. By using the head's physical structure (interaural time differences and head shadowing) as the filtering mechanism, the system eliminates the need for computationally intensive VAD and adaptive filter weight optimization, significantly reducing processing requirements.
Solution Approach 2:
The system uses the head's own acoustic properties to provide directionality. The natural interaural time and level differences created by the head's geometry serve as the filtering mechanism, making the system self-sufficient without requiring external complex processing algorithms or VAD assistance.
3Adaptability or versatility
If BSS schemes are used, then directional responses are produced in all spatial locations, but computational overload and dependence on reverberation increase
Solution Approach 1:
The patent extracts only the necessary spatial information from the microphone signals by directly exploiting the head's acoustic shadowing and interaural level differences. Rather than computing complete BSS decompositions that separate all sound sources in all spatial locations, the system focuses on extracting directional information for the desired sound source using the head's natural acoustic properties, significantly reducing computational load.
Solution Approach 2:
Instead of attempting to separate all sound sources in all spatial directions (excessive action), the patent applies a more targeted approach using partial beamforming that focuses on the desired direction and exploits only the necessary interaural differences, avoiding the computational overload of complete BSS while still achieving adequate spatial separation.
4Reliability
If VAD are used in LMS systems, then desired signal is captured in high signal-to-noise ratio environments, but performance degrades significantly in low signal-to-noise ratio environments
Solution Approach 1:
The system uses the head's own acoustic properties to provide robust directionality that works independently of the signal-to-noise ratio. The interaural time and level differences are physical phenomena that exist regardless of signal strength, making the system self-sufficient and adaptable to all listening conditions without requiring VAD assistance.
Solution Approach 2:
The patent prepares for low signal-to-noise ratio conditions by using the head's natural acoustic filtering, which provides built-in noise rejection capability. The head shadowing and interaural level differences create a protective effect that cushions against noise before it can interfere with the desired signal, enabling reliable operation in challenging acoustic environments.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances sound intelligibility and directional signal production in complex environments, providing superior noise rejection and operating effectively across varying signal-to-noise ratios, including low-frequency and high-frequency sound cancellation.
Implementation Method 1
The interaural time difference, arising from the spacing between microphones on each side of the head (ranging from 18 to 28 cm), can be used to cancel relatively low frequency sounds, depending on the direction of arrival
Implementation Method 2
the head shadowing provides a natural level suppression of contralateral sounds (i.e. sounds presented from each side of the head)
Implementation Method 3
the interaural level difference (ranging from 0 to 18 dB), can be used to cancel high frequency sounds depending on their direction of arrival in a weighted sum configuration
Data Source
AI summary
A system and method of producing a directional output signal is described including the steps of: detecting sounds at the left and rights sides of a person's head to produce left and right signals; determining the similarity of the signals; modifying the signals based on their similarity; and combining the modified left and right signals to produce an output signal.