Acoustic Feedback Detection via Spectral Power Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional acoustic feedback detection methods face issues with missing detection and false detection due to limitations in wide-band filtering and reliance on signal waveform periodicity, which affect the accuracy and reliability of feedback detection.
Innovation Solution
The proposed method performs time-frequency conversion on a received time-domain signal to determine power peak values and calculate power sum and average power values, using these to assess judgment values and thresholds across frequency bands, thereby determining the occurrence of acoustic feedback without relying on band-pass filters or waveform periodicity analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If multiple band-pass filters are used to extract signals in predetermined frequency bands, then the detection process can be simplified, but the full frequency band cannot be covered due to limitations in center frequency and bandwidth, causing missing detection
Solution Approach 1:
The frequency band is divided into multiple sub-bands using band-pass filters with predetermined center frequencies and bandwidths. This segmentation allows the detection process to be simplified while systematically covering the entire frequency range through multiple discrete frequency points, resolving the contradiction between operational simplicity and comprehensive frequency coverage.
2Ease of operation
If acoustic feedback detection is based on judging whether the signal waveform is sinusoidal periodic, then the detection method is straightforward, but false detection may occur
Solution Approach 1:
The patent replaces the mechanical waveform shape judgment method with a spectral analysis approach using power spectrum density estimation. Instead of visually or computationally checking for sinusoidal periodicity, the system transforms the time-domain signal into the frequency domain and identifies feedback based on spectral characteristics, thereby eliminating false detections while maintaining detection simplicity.
3Area of stationary object
If a wide-band filter is used to process the sound signal, then the entire frequency range can be analyzed, but the problem of missing detection and false detection occurs due to reliance on waveform periodicity
Solution Approach 1:
The patent transitions from time-domain waveform analysis to frequency-domain spectral analysis, adding a dimensional transformation to the detection process. By estimating power spectrum density and analyzing frequency components rather than examining time-domain periodicity, the system achieves both comprehensive frequency range coverage and improved detection precision, resolving the contradiction between wide frequency analysis and accurate feedback detection.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed are a sound feedback detection method and device. The method comprises: performing time frequency conversion on a received time domain signal to obtain a frequency domain signal; calculating a power sum value of a plurality of adjacent points on the left and right of a power peak value and an average power value of the frequency signal; determining a judgement value according to the power sum value and the average power value; determining a corresponding pre-set first threshold value according to a frequency band range into which the power peak value frequency falls, and if the judgement value is greater than the pre-set first threshold value, determining that the judgement value is a judgement value to be counted corresponding to the frequency band range; counting the number of the judgement values to be counted corresponding to the frequency band range within the pre-set time, and a repetition duration of the power peak value which falls into the frequency band range within the pre-set time; and if the number is greater than a pre-set second threshold value or the repetition duration is greater than a pre-set third threshold value, determining that sound feedback is occurring. In this way, the technical solution can improve the accuracy and reliability of sound feedback detection.