Voice Denoising with Interruption-Aware Algorithm Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice data processing systems in vehicles face challenges in effectively distinguishing and filtering out noise from human voice data, particularly in scenarios involving synthetic voice interruptions, leading to suboptimal voice quality and recognition accuracy.
Innovation Solution
The method involves invoking different denoising algorithms based on the presence of a synthetic voice interruption signal, with one algorithm highlighting original voice features when no interruption is detected and another highlighting human voice features when interruptions occur, thereby maintaining voice quality and improving recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single denoising algorithm is used to process all voice data, then the processing flow is simple, but the voice quality and recognition accuracy deteriorate in scenarios with synthetic voice interruptions
Solution Approach 1:
The patent segments the voice data processing into two distinct paths based on the presence of synthetic voice interruptions. A first denoising algorithm processes normal voice data, while a second denoising algorithm processes data with synthetic voice interruptions. This segmentation allows each algorithm to be optimized for its specific scenario, improving voice quality and recognition accuracy without significantly increasing overall system complexity.
Solution Approach 2:
The patent implements dynamic switching between different denoising algorithms based on the detection of synthetic voice interruptions. The system dynamically adjusts the processing approach by detecting interruption signals and selecting the appropriate algorithm, allowing the processing flow to adapt to changing conditions while maintaining reasonable complexity.
2Manufacturing precision
If different denoising algorithms are used for different scenarios, then voice quality and recognition accuracy improve, but the device complexity increases
Solution Approach 1:
The patent introduces an intermediary component - the detection module that identifies synthetic voice interruption signals. This intermediary acts as a gateway that routes voice data to the appropriate denoising algorithm, managing the complexity of having multiple algorithms while maintaining the benefits of scenario-specific processing.
Solution Approach 2:
The patent changes the processing parameters by selecting different algorithms based on the presence of interruption signals. This parameter-based approach allows the system to maintain simplicity in the processing flow structure while achieving improved voice quality through parameter-driven algorithm selection.
3Measurement precision
If complex denoising processing is applied to all voice data, then recognition accuracy improves, but computational time delay increases
Solution Approach 1:
The patent applies partial action by using a simpler first denoising algorithm for normal voice data that does not contain synthetic voice interruptions. Only when interruption signals are detected does the system apply the more complex second denoising algorithm. This partial application of complex processing reduces overall computational time delay while maintaining recognition accuracy where it is most needed.
Data Source
AI summary
The present disclosure provides a method for denoising voice data, an electronic device, and a computer readable storage medium. The present disclosure relates to the technical field of artificial intelligence, such as Internet of Vehicles, smart cockpit, smart voice, and voice recognition. A specific embodiment of the method includes: receiving an input to-be-played first piece of voice data; and invoking, in response to not detecting a synthetic voice interruption signal in a process of playing the first piece of voice data, a preset first denoising algorithm to filter out noise data except for the first piece of voice data.