Voice Noise Classification Using Frequency Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional electronic devices face difficulties in accurately distinguishing between voice and noise intervals, particularly in cases of non-stationary noise and low signal-to-noise ratio (SNR) signals, which affects call quality and voice recognition.
Innovation Solution
An electronic device equipped with multiple microphones, where a processor converts signals from these microphones into frequency signals and determines voice or noise based on frequency-related correlations, energy differences, and spectral variances, using methods like magnitude squared coherence (MSC) to improve classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional single-microphone detection methods are used, then device complexity is low, but measurement precision of voice and noise detection deteriorates
Solution Approach 1:
The patent divides the audio signal detection task into multiple independent microphone channels, where each microphone captures audio from a different spatial position. The processor then segments the analysis by evaluating frequency-related correlations specifically between corresponding frequency components of multiple microphones, allowing precise voice-noise differentiation while maintaining manageable device complexity through modular signal processing.
Solution Approach 2:
The patent transitions from single-microphone temporal analysis to multi-microphone spatial-frequency analysis. By introducing the spatial dimension through multiple microphones and analyzing frequency-related correlations across this new dimension, the system achieves superior detection accuracy for non-stationary noise and low SNR signals that cannot be resolved by conventional single-channel methods.
2Measurement precision
If conventional energy-based detection is used, then processing simplicity is maintained, but measurement precision deteriorates for non-stationary noise and low SNR signals
Solution Approach 1:
The patent moves beyond simple energy detection by transforming the audio signal into the frequency domain and analyzing frequency-related correlations. This parameter change from temporal energy to spectral correlation enables accurate detection of non-stationary noise and low SNR signals, as frequency-domain analysis captures the characteristic spectral patterns that distinguish voice from noise even when energy levels are low or rapidly changing.
3Reliability
If single-microphone detection is used, then device complexity is low, but reliability of voice and noise classification deteriorates in low SNR conditions
Solution Approach 1:
The patent merges the signals from multiple microphones by evaluating frequency-related correlations between them. This combining approach leverages the spatial diversity of multiple microphones to extract robust voice characteristics even in low SNR conditions, where individual microphone signals may be unreliable. The correlation-based merging enhances reliability by identifying consistent spectral patterns across multiple spatial channels.
Data Source
AI summary
An electronic device includes a first microphone that receives a sound generated for a specific time period, from the outside, a second microphone, which is disposed at a location spaced apart from the first microphone and which receives the sound, an audio converter comprising audio converting circuitry, and a processor electrically connected with the first microphone, the second microphone, and the audio converter. The processor is configured to convert the sound obtained from the first microphone, into a first signal and to convert the sound obtained from the second microphone, into a second signal, using the audio converter, and to determine the sound, which is generated for the specific time period, as a voice or a noise based on a frequency-related correlation between the first signal and the second signal.