Bone and Air Conduction Voice Mixing for Adaptive Noise Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio signal processing methods using both air and bone conduction sensors either fail to effectively combine signals in varying noise environments, leading to increased ambient noise or muffled voice, or require complex noise estimation that increases computational load and latency.
Innovation Solution
An adaptive audio signal processing method that combines bone-conducted and air-conducted signals by performing spectral analysis using FFT or similar techniques to determine a cutoff frequency, allowing for quick adaptation to noise conditions without relying on noise estimation, and using filter banks to mix the signals based on this frequency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If only air-conducted signal is used in the output signal, then the signal can capture full spectral bandwidth, but the output signal will contain more ambient noise
Solution Approach 1:
The patent combines air-conducted and bone-conducted signals using adaptive filtering techniques. The bone-conducted signal serves as a reference to estimate and remove ambient noise from the air-conducted signal, producing an output that retains full spectral bandwidth while reducing noise. This merging of two signal paths resolves the contradiction between maintaining information completeness and reducing harmful noise.
2Object-affected harmful factors
If only bone-conducted signal is used in the output signal, then the ambient noise is reduced, but the voice signal will be strongly low-pass filtered causing muffled sound
Solution Approach 1:
The patent uses the bone-conducted signal as an intermediary reference to extract noise characteristics without directly using it as the output. The adaptive filter uses this reference to subtract noise from the air-conducted signal, allowing the output to maintain the full frequency range of the air-conducted signal while removing noise. This intermediary approach avoids the muffled sound problem while still achieving noise reduction.
3Device complexity
If static mixing scheme is used to combine bone-conducted and air-conducted signals, then the processing is simple, but the mixing is not adaptive to varying noise environments
Solution Approach 1:
The patent implements dynamic adaptation by continuously updating filter coefficients based on the statistical properties of the input signals. The adaptive algorithm adjusts the mixing ratio between bone-conducted and air-conducted signals in real-time according to the current noise environment. This dynamic approach maintains relatively simple processing while achieving environment-specific optimization, resolving the contradiction between complexity and adaptability.
4Adaptability or versatility
If adaptive noise estimation is used to mix signals based on estimated noise, then the mixing adapts to noise conditions, but the noise estimators are slow and inaccurate with increased computational complexity
Solution Approach 1:
The patent extracts noise characteristics directly from the bone-conducted signal without requiring complex noise estimation algorithms. Since the bone-conducted signal contains primarily voice information with minimal ambient noise, it serves as a clean reference for extracting noise components present in the air-conducted signal. This extraction approach achieves adaptive noise cancellation with reduced computational complexity compared to traditional noise estimation methods.
Data Source
AI summary
Disclosed is an audio signal processing method, including measuring a voice signal by internal and external sensors. The internal sensor measures voice signals that propagate internally to the user's head. The external sensor measures voice signals that propagate externally to the user's head. The internal and external sensors produces first and second audio signals, respectively. The method further includes: processing the first audio signal to produce a first audio spectrum on a frequency band; processing the second audio signal to produce a second audio spectrum on the frequency band; computing a first cumulated audio spectrum by cumulating first audio spectrum values; computing a second cumulated audio spectrum by cumulating second audio spectrum values; determining a cutoff frequency by comparing the first and second cumulated audio spectra; and producing an output signal by combining the first audio signal and the second audio signal based on the cutoff frequency.