Smart Noise Reduction Using Pattern Recognition for Unsteady Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise reduction methods in intelligent devices are ineffective against unsteady noise, leading to speech distortion and reduced speech recognition rates, and existing air conditioning systems fail to adequately suppress operational noise.
Innovation Solution
A smart noise reduction device utilizing a control device with a noise pattern recognizer, database, and filter to destructively interfere noise signals, and a heart sound sensing device with a heart sound pattern recognizer and filter to eliminate noise through destructive interference of sound waves.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If spectral subtraction method is used to eliminate noise, then noise elimination is achieved, but speech distortion occurs and speech recognition rate declines
Solution Approach 1:
The patent changes the parameters of noise elimination by using different filtering methods (spectral subtraction, Wiener filtering, adaptive noise cancellation) with adjustable parameters. The system dynamically adjusts filtering parameters based on noise characteristics to eliminate noise while preserving speech quality and recognition accuracy.
2Manufacturing precision
If Wiener filtering method is used to suppress noise, then speech distortion is reduced, but noise reduction performance deteriorates in unsteady noise with large changes
Solution Approach 1:
The patent implements dynamic noise reduction by continuously adapting the filtering method and parameters based on real-time noise characteristics. The system switches between different noise reduction algorithms and adjusts their parameters dynamically to handle unsteady noise with large changes while maintaining speech quality.
3Object-affected harmful factors
If adaptive noise cancellation method is used with directional microphone, then ambient noise is reduced, but device complexity increases
Solution Approach 1:
The patent makes the noise reduction system universal by implementing multiple noise reduction algorithms (spectral subtraction, Wiener filtering, adaptive noise cancellation) within a single device. The system can adaptively select and combine these methods based on different noise scenarios, providing comprehensive noise reduction without requiring multiple specialized devices.
4Object-affected harmful factors
If noise reduction filtering is applied to heart sound signals, then noise is suppressed, but signal-to-noise ratio improvement is limited without precise noise pattern matching
Solution Approach 1:
The patent uses noise pattern recognition to create accurate copies or representations of noise signals. By identifying and copying noise patterns, the system can generate precise inverse noise signals for destructive interference, significantly improving the signal-to-noise ratio in heart sound detection.
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
Effectively suppresses noise from various sources and improves speech recognition and heart sound interpretation by using destructive interference of noise patterns, enhancing signal-to-noise ratio and accuracy.
Implementation Method 1
The mixed audio signal includes an inverse noise of ambient noise. When the input noise pattern/waveform signal (inverse noise) and the noise source enter the ear canal of human together, the inverted noise and the noise source will produce destructive interference because of mutual cancellation of the waveform, so as to achieve the performance of reducing the noise decibel.
Data Source
AI summary
The present invention discloses a smart noise reduction device including a control device; an audio waveform pattern recognizer coupled to the control device for identifying an audio mixed signal including a regularity signal and a non-regularity signal; an audio waveform pattern database coupled to the control device, including at least one audio type, each having a plurality of preset second regularity signals; and an audio filter coupled to the control device to obtain the regularity signal.