Wake-Up Word Recognition Model Inappropriate Content Screening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user terminals face challenges in implementing large vocabulary continuous speech recognition (LVCSR) systems due to high computation and memory requirements, making it difficult to accurately process voice inputs and detect inappropriate words.
Innovation Solution
A user terminal is equipped with a communication interface, memory, microphone, speaker, display, main processor, and a sub-processor that activates the main processor by recognizing a wake-up word in a voice input. The system determines whether the wake-up word input includes a specified inappropriate word, preventing its registration and ensuring appropriate voice input processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LVCSR system is implemented in user terminal, then speech recognition accuracy is improved, but computation and memory requirements increase significantly
Solution Approach 1:
The speech recognition system is divided into two parts: a lightweight wake-up word recognition model running on the user terminal for initial activation, and a full LVCSR system running on a server for accurate speech processing. This segmentation allows the terminal to perform basic functions with limited resources while maintaining the option for high-accuracy recognition when connected to external services.
2Device complexity
If LVCSR system is not implemented in user terminal, then device complexity is reduced, but ability to detect inappropriate words is worsened
Solution Approach 1:
A keyword detection module serves as an intermediary component in the user terminal that screens voice inputs for inappropriate content before processing. This module uses a predefined keyword list to identify potentially problematic words, providing basic content filtering capability without requiring the full computational power of an LVCSR system.
3Adaptability or versatility
If wake-up word recognition model is trained with inappropriate voice input, then model adaptability increases, but inappropriate word registration occurs
Solution Approach 1:
The system performs preliminary detection of inappropriate words using the keyword detection module before the wake-up word recognition model is trained. By screening the training voice input in advance and rejecting any containing inappropriate keywords, the system prevents harmful content from being incorporated into the model while still allowing adaptable training with appropriate content.
Data Source
AI summary
Disclosed is an electronic device including a communication interface, a memory, a microphone, a speaker, a display, a main processor, and a sub-processor activating the main processor by recognizing a wake-up word included in a voice input. The at least one memory stores instructions that, when executed, cause the main processor to receive a first voice input to register the wake-up word, when the first voice input does not include a specified word, to receive a second voice input including a word identical to the first voice input, through the microphone, to generate a wake-up word recognition model for recognizing the wake-up word, and to store the generated wake-up word recognition model in the at least one memory, and when the first voice input includes the specified word, to output information for requesting a third voice input, through the speaker or the display.