Noise Signal Suppression via Independent Peak Spectrum Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise suppression methods struggle to detect and differentiate special signals like siren sounds and notification sounds in real-time due to their similarity to speech signals, leading to delayed processing and increased signal load, which can result in incorrect identification and deterioration of speech quality.
Innovation Solution
A device and method that convert sound signals into the frequency domain, extract independent peak spectra, and determine their persistence to distinguish special noise signals from speech, allowing for real-time detection and suppression without requiring extensive pattern analysis or harmonic analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional noise suppression methods (spectrum subtraction) are used to detect and suppress noise signals, then stationary noise signals can be suppressed effectively, but special signals like siren sounds and notification sounds cannot be distinguished from speech signals, leading to incorrect identification and deterioration of speech quality
Solution Approach 1:
The patent changes the detection parameters from spectrum-based analysis to zero-crossing rate and signal strength analysis. By using zero-crossing rate (the number of times the signal crosses the zero axis per unit time) and signal strength thresholds, the system can distinguish special signals from speech without requiring complex pattern matching, thereby improving both reliability and adaptability
Solution Approach 2:
The patent replaces the conventional spectrum subtraction method (which relies on frequency domain analysis) with a time-domain based method using zero-crossing rate detection. This substitution allows for simpler, faster detection that works effectively for both stationary noise and special signals like sirens and notification sounds
2Measurement precision
If extensive pattern analysis and harmonic analysis are performed to distinguish special signals from speech, then detection accuracy improves, but processing time increases and real-time detection becomes difficult
Solution Approach 1:
The patent extracts only the essential characteristics (zero-crossing rate and signal strength) needed for special signal detection, discarding the need for extensive pattern analysis and harmonic analysis. This extraction of key features enables fast, real-time detection while maintaining sufficient accuracy to distinguish special signals from speech
Solution Approach 2:
The patent uses a simplified detection approach that performs partial analysis (zero-crossing rate and signal strength only) rather than complete spectral analysis. This partial action is sufficient for detecting special signals in real-time without the computational burden of full pattern matching
3Reliability
If the analysis duration is extended to several tens or hundreds of milliseconds to accurately detect special signals, then detection reliability improves, but the processing delay increases beyond acceptable limits for mobile communications
Solution Approach 1:
The patent uses the inherent characteristics of special signals (high zero-crossing rate and specific signal strength patterns) to enable self-identification within short time windows. The detection method does not require long-term accumulation of data because the statistical properties of special signals manifest quickly, allowing reliable detection at processing speeds suitable for mobile communications
Data Source
AI summary
Provided is a noise signal suppressing device including: an input unit configured to receive a sound signal; a time/frequency converting unit; an independent peak spectrum extracting unit configured to extract a peak spectrum having independence; a persistence determining unit configured to determine that the peak spectrum having independence persists for a predetermined period or longer; a noise-signal suppressing unit configured to suppress the peak spectrum having independence as the noise signal. The independent peak spectrum extracting unit includes: a first peak extracting unit configured to extract a peak spectrum having higher energy than that of an adjacent frequency signal, and a second peak extracting unit configured to extract a peak spectrum maintaining a frequency interval of equal to or larger than a predetermined value with respect to a peak spectrum adjacent thereto as the peak spectrum having independence.