Audio Howling Control via Neural Network Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Audio systems, such as those in vehicles and public announcement systems, experience howling due to acoustic coupling between microphones and speakers, leading to unpleasant feedback loops, which existing technologies have not effectively addressed.
Innovation Solution
An audio system comprising a speaker, a microphone, a mixer module, a filter module, and a detector module, where the detector module uses a deep neural network to identify howling frequencies and adjusts a notch filter to attenuate them, thereby reducing the magnitude of the affected signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If acoustic coupling between microphone and speaker is present, then the audio system can function with feedback, but howling occurs as a harmful effect
Solution Approach 1:
The detector module uses the feedback signal itself (the harmful howling) as the detection target, identifying its frequency characteristics and converting this harmful acoustic coupling into useful information for filtering. By detecting the howling frequency from the microphone signal and applying targeted filtering, the system transforms the problematic feedback into a detectable pattern that can be eliminated.
Solution Approach 2:
The filter module extracts and removes the specific howling frequency component from the mixed audio signal. By identifying the narrow frequency band where howling occurs and applying a notch filter at that specific frequency, the system separates and eliminates only the harmful feedback component while preserving the rest of the audio signal.
2Ease of operation
If feedback loop is established between speaker and microphone, then system can operate, but unpleasant howling sound is generated
Solution Approach 1:
The system uses the microphone to continuously monitor the speaker output and feeds this information back to the detector module. This feedback loop, which would normally cause howling, is instead utilized to detect the presence and frequency of howling in real-time, enabling dynamic adjustment of the filter to suppress the harmful sound while maintaining system operation.
3Object-affected harmful factors
If notch filter is applied to reduce howling frequency, then howling is minimized, but frequency response may be affected
Solution Approach 1:
The filter module applies a narrow notch filter that affects only the specific howling frequency and its immediate vicinity, while leaving the rest of the frequency spectrum unchanged. This localized filtering approach minimizes the impact on overall frequency response by concentrating the filtering action only where the harmful howling occurs, rather than applying broad frequency attenuation.
Solution Approach 2:
The system dynamically adjusts the notch filter parameters (center frequency and bandwidth) based on the detected howling frequency. Rather than using a fixed filter, the filter characteristics change in real-time to track and suppress the varying howling frequency, maintaining frequency response accuracy for non-howling frequencies while effectively reducing howling.
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 minimizes howling frequencies by accurately detecting and canceling them, improving the audio quality in vehicles and other systems with microphone and speaker setups.
Implementation Method 1
Howling is an unpleasant sound that occurs when there is acoustic coupling between a microphone and a loudspeaker system. Howling may refer to the situation where output from the speakers is fed back into the system via the microphone
Implementation Method 2
the filter module is configured to decrease a magnitude of the filtered signal at the howling frequency
Data Source
AI summary
An audio system includes: a speaker; a microphone that generates a microphone signal based on sound output from the speaker; a mixer module configured to generate a mixed signal by mixing the microphone signal with an audio signal; a filter module configured to filter the mixed signal to produce a filtered signal and to apply the filtered signal to the speaker; and a detector module configured to determine a howling frequency in the microphone signal attributable to sound output from the speaker, where the filter module is configured to decrease a magnitude of the filtered signal at the howling frequency.


