Adaptive Voice Recognition Sensitivity via Noise-Sensitivity Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Voice recognition devices face challenges in accurately recognizing user voices in various noise environments due to varying noise levels, leading to inconsistent voice recognition rates.
Innovation Solution
A method and apparatus that acquire a noise level in the environment using microphones, input this noise level into a previously learned noise-sensitivity model to determine an optimum sensitivity, and use this sensitivity to recognize user voices. This involves learning a noise-sensitivity model in multiple noise environments and configuring noise-sensitivity data sets based on recognition rates, allowing for adaptive voice recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the voice recognition device uses a fixed sensitivity setting, then the device structure is simple, but the voice recognition rate becomes inconsistent in various noise environments
Solution Approach 1:
The noise-sensitivity model is learned in advance through training in multiple noise environments with different noise levels. This pre-trained model stores optimal sensitivity parameters for various noise conditions, allowing the device to quickly adapt to current environmental noise without complex real-time adjustments, thereby improving recognition reliability while maintaining relatively simple device structure
Solution Approach 2:
The voice recognition sensitivity is dynamically adjusted based on the current noise level detected by the microphone. The system continuously monitors environmental noise and adapts the sensitivity parameter in real-time according to the pre-trained noise-sensitivity model, enabling consistent recognition performance across varying noise conditions
2Measurement precision
If the voice recognition device increases sensitivity to capture weak voices, then weak voices can be recognized, but noise interference increases leading to false recognition
Solution Approach 1:
The system changes the sensitivity parameter dynamically based on the detected noise level. When noise is high, sensitivity is reduced to avoid false recognition; when noise is low, sensitivity is increased to capture weak voices. This parameter adaptation is guided by the pre-trained noise-sensitivity model that learned optimal settings across various noise conditions
Solution Approach 2:
The system implements feedback by continuously monitoring the current noise level through the microphone and adjusting the sensitivity parameter accordingly. The noise level detection result feeds back to the sensitivity adjustment mechanism, creating a closed-loop system that adapts to environmental conditions in real-time
3Reliability
If the voice recognition device decreases sensitivity to reduce noise interference, then false recognition decreases, but weak voices cannot be recognized
Solution Approach 1:
The sensitivity parameter is dynamically adjusted based on real-time noise level detection. The system transitions between different sensitivity states according to environmental conditions, enabling it to maintain high recognition accuracy in noisy environments while preserving the ability to detect weak voices in quiet environments
Solution Approach 2:
The noise-sensitivity model is pre-trained with optimal sensitivity parameters for various noise levels and environments. This preliminary training allows the system to quickly select appropriate sensitivity settings without trial-and-error adjustments, ensuring both accurate noise rejection and weak voice detection capability
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
The solution enables accurate voice recognition by setting optimal sensitivity for the current noise environment, improving recognition rates and reliability across different noise levels.
Implementation Method 1
acquiring a microphone detection signal through at least one microphone of the voice recognizing apparatus
Data Source
AI summary
Disclosed herein is a method for intelligently recognizing voice by a voice recognizing apparatus in various noise environments. The method includes acquiring a first noise level for an environment in which the voice recognizing apparatus is located, inputting the first noise level into a previously learned noise-sensitivity model to acquire a first optimum sensitivity, and recognizing a user's voice based on the first optimum sensitivity. The noise-sensitivity model is learned in a plurality of noise environments acquiring different noise levels, so that it is possible to accurately acquire an optimum sensitivity corresponding to a noise level depending on an operating state when an IoT device (voice recognizing apparatus) is in operation. At least one of the voice recognizing apparatus and an intelligent computing device of present disclosure can be associated with artificial intelligence modules, drones (unmanned aerial vehicles (UAVs)), robots, augmented reality (AR) devices, virtual reality (VR) devices, devices related to 5G service, etc.