Wearable Device Noise Reduction via Harmonic Filter Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Beamforming in wearable electronic devices often requires high gain at low frequencies to compensate for low frequency roll-off, which degrades the signal-to-noise ratio due to amplified noise.
Innovation Solution
A method using a processor to generate a beamformed signal and estimate a fundamental frequency, configuring a filter with passbands and stopbands to suppress noise signals while passing voiced speech signals, thereby improving the signal-to-noise ratio.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If beamforming is used to focus on the wearer's mouth, then spatial filtering capability is improved, but low frequency roll-off occurs requiring high gain equalization which amplifies noise
Solution Approach 1:
The frequency spectrum is segmented into multiple bands (low frequency band below first harmonic, mid frequency band between harmonics, high frequency band above second harmonic). Different filtering strategies are applied to each band: high gain equalization is applied to the low frequency band to compensate for beamforming roll-off, while band-stop filtering is applied to the mid frequency band to suppress noise without amplification. This segmentation allows simultaneous optimization of low frequency response and noise suppression.
Solution Approach 2:
Different quality characteristics are applied to different frequency regions. The low frequency region receives high gain equalization to compensate for beamforming attenuation, while the mid frequency region receives aggressive noise suppression through band-stop filters. This local differentiation of filtering characteristics optimizes the overall signal-to-noise ratio by treating each frequency region according to its specific requirements.
2Use of energy by stationary object
If high gain equalization is applied to compensate for low frequency roll-off, then low frequency response is improved, but signal-to-noise ratio deteriorates due to noise amplification
Solution Approach 1:
The frequency spectrum is divided into distinct segments with different processing strategies. The low frequency segment (below first harmonic) receives high gain equalization to restore beamforming attenuation, while the mid frequency segment (between harmonics) receives band-stop filtering to suppress noise without high gain amplification. This segmentation resolves the contradiction by applying high gain only where necessary for frequency response compensation.
Solution Approach 2:
The harmonic structure of voiced speech, which could be seen as a constraint, is converted into a beneficial feature. Band-stop filters are placed at frequencies between harmonics where speech energy is naturally low but noise may be present. This approach uses the predictable harmonic structure to identify optimal noise suppression frequencies that do not adversely affect speech quality.
3Object-affected harmful factors
If band-stop filtering is applied to suppress noise between harmonic frequencies, then signal-to-noise ratio is improved, but voiced speech may be attenuated if filtering is too aggressive
Solution Approach 1:
Different filtering strengths are applied to different frequency regions. Band-stop filters with high attenuation are applied only to the mid frequency band between harmonics where speech energy is naturally low. The low frequency band below the first harmonic receives high gain equalization without band-stop filtering, preserving voiced speech components. This localized application of aggressive filtering minimizes impact on speech quality while maximizing noise suppression.
Solution Approach 2:
The filtering characteristics are made adaptive based on the estimated fundamental frequency and harmonic structure. As the fundamental frequency changes with different speech content, the filter frequencies and bandwidths are dynamically adjusted to track the harmonic structure. This dynamic adaptation ensures that noise suppression remains effective across varying speech conditions while preserving speech quality.
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
The solution effectively suppresses noise signals without cutting off voiced speech, resulting in a comparative improvement of the signal-to-noise ratio and reducing noise amplification at low frequencies.
Implementation Method 1
a first electro-acoustic input transducer and a second electro-acoustic input transducer arranged to pick up a first acoustic signal and convert the first acoustic signal to a first microphone signal and a second microphone signal
Implementation Method 2
configuring a first filter with one or more first passbands, including an upper first passband, at one or more integer multiples of the first frequency value; and one or more first stop bands adjacent the one or more passbands
Data Source
AI summary
A method at a wearable electronic device with: a first electro-acoustic input transducer and a second electro-acoustic input transducer arranged to pick up a first acoustic signal and convert the first acoustic signal to a first microphone signal and second microphone signal; and a third electro-acoustic input transducer arranged to pick up a second acoustic signal and convert the second acoustic signal to a third microphone signal; and a processor (140). The method comprises: generating a beamformed signal based on the first microphone signal (x1) and the second microphone signal; estimating a first frequency value representing a fundamental frequency in one or more of: the first microphone signal, the second microphone signal and the third microphone signal; configuring a first filter with one or more passbands at one or more integer multiples of the first frequency value and one or more stop bands adjacent the one or more stop bands; and filtering, using the first filter, one or more of: the first microphone signal, the second microphone signal and the beamformed signal.