Asymmetric Two-Microphone Noise Reduction via Energy Transfer Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise reduction methods for communications devices, such as mobile phones, fail to robustly improve communication quality across a wide range of ambient noise conditions, often degrading voice quality or requiring noise-free calibration.
Innovation Solution
A method and apparatus using asymmetric two-microphone techniques, where one microphone is positioned near the user's mouth and the other farther away, calculate energy transfer functions to generate a noise-reduced output signal by attenuating noise components while maintaining voice signal integrity, using pre-processing to subtract pseudo noise signals and avoid feature detection complications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If single-microphone noise reduction methods are used to reduce noise level, then noise suppression is achieved, but voice quality degrades under high noise conditions
Solution Approach 1:
The patent divides the noise reduction task into two segments: first using a single-microphone method for initial noise suppression, then applying a two-microphone method for further noise reduction. This segmented approach allows each method to operate in its optimal range, preventing voice quality degradation that would occur if either method operated alone under high noise conditions.
Solution Approach 2:
The patent merges the output of the single-microphone noise reduction method with the two-microphone noise reduction method. By combining these methods, the system achieves superior noise suppression while maintaining voice quality, as each method compensates for the other's limitations under different noise conditions.
2Object-affected harmful factors
If asymmetric two-microphone methods are used to improve noise reduction, then noise suppression improves, but the system requires complex feature detection and calibration
Solution Approach 1:
The patent performs preliminary action by using the single-microphone noise reduction method first to suppress dominant noise components before applying the two-microphone method. This preliminary noise reduction simplifies the subsequent feature detection task, reducing complexity while maintaining effective noise suppression.
Solution Approach 2:
The patent applies partial action by using the two-microphone method specifically for residual noise reduction after the single-microphone method has already addressed the dominant noise. This partial application of the more complex method reduces the burden of feature detection while achieving comprehensive noise suppression.
3Device complexity
If simple subtraction of noise microphone signal is used, then device complexity is reduced, but noise reduction effectiveness is poor due to noise variation
Solution Approach 1:
The patent implements feedback by using the output of the single-microphone noise reduction method as input to the two-microphone method. This feedback loop allows the system to adapt to varying noise conditions, improving noise reduction effectiveness while maintaining reasonable processing complexity through the structured feedback architecture.
Data Source
AI summary
A method and apparatus are provided for generating a noise reduced output signal from sound received by a first microphone. The method includes transforming the sound received by the first microphone into a first input signal and transforming sound received by a second microphone into a second input signal. The method includes calculating, for each of a plurality of frequency components, an energy transfer function value as a real-valued quotient by dividing a temporally averaged product of an amplitude of the first input signal and the second input signal by a temporally averaged absolute square of the second input signal, calculating a gain value as a function of the calculated energy transfer function value, and generating the noise reduced output signal based on the product of the first input signal and the calculated gain value at each of the plurality of frequency components.