AI Noise Removal in Electronic Apparatus Audio Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise cancellation technologies face challenges in completely removing noise signals and suffer from low accuracy in identifying noise signals, especially when input data differs from the training data.
Innovation Solution
An electronic apparatus that uses a gain value associated with the signal-to-noise ratio between a voice signal and a noise signal to remove noise from an audio signal. This apparatus includes a processor that converts audio signals between time and frequency domains, obtains gain values through filtering and neural network processing, and applies these gain values to remove noise from the audio signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If noise cancellation is performed by generating an opposite noise signal, then noise removal is attempted, but complete noise cancellation is difficult to achieve
Solution Approach 1:
The patent replaces traditional mechanical noise cancellation methods (generating opposite noise signals) with an artificial intelligence-based signal processing system. The AI model analyzes the audio signal in the frequency domain, identifies noise components through spectral analysis, and selectively removes them without requiring precise generation of opposite noise signals, thereby achieving more accurate noise removal.
Solution Approach 2:
The patent transforms the audio signal from time domain to frequency domain using Short-Time Fourier Transform (STFT), changing the representation parameters of the signal. This parameter transformation enables more effective noise identification and separation by analyzing frequency spectral characteristics, allowing the system to distinguish noise from speech based on frequency distribution patterns.
2Object-affected harmful factors
If artificial intelligence model is used to remove noise, then noise removal function is provided, but accuracy decreases when input data differs from training data
Solution Approach 1:
The patent performs preliminary processing of the audio signal by transforming it to the frequency domain and extracting spectral features before inputting to the AI model. This preliminary action prepares the data in an optimized format that enhances the model's ability to generalize across different noise types and conditions, improving adaptability without requiring retraining.
Solution Approach 2:
The patent introduces an intermediate signal processing stage using Short-Time Fourier Transform as a mediator between the raw audio input and the AI model. This intermediary transformation converts time-domain signals into frequency-domain representations, creating a standardized intermediate format that improves the model's robustness to variations in input data characteristics.
3Object-affected harmful factors
If noise signal is removed using traditional methods, then some noise reduction is achieved, but communication quality remains insufficient
Solution Approach 1:
The patent replaces traditional mechanical noise cancellation approaches with an AI-based intelligent system that can adaptively analyze and remove noise while preserving speech quality. The system uses neural networks to learn optimal noise removal strategies, achieving more reliable communication quality by intelligently distinguishing between noise and speech components rather than applying fixed mechanical filtering methods.
Data Source
AI summary
An example electronic apparatus includes a memory configured to store at least one instruction and at least one processor connected to the memory to control the electronic apparatus. The at least one processor is configured to, by executing the at least one instruction, obtain a first audio signal including a voice signal and a noise signal, convert the first audio signal in a time domain to a second audio signal in a frequency domain, obtain a first gain value representing a Signal-to-Noise Ratio (SNR) from the second audio signal, obtain a second gain value with a first dynamic range by filtering the first gain value, obtain a third gain value by inputting the second gain value to a neural network model trained to output a signal from which noise is removed, and convert the second audio signal to a third audio signal from which at least a portion of the noise signal is removed, using the third gain value.