Speech Verification Model Using Merged Recognition and Language Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech recognition technologies face challenges in accurately evaluating linguistic accuracy and confidence measures, often resulting in ungrammatical sentences due to limited feature consideration, and are complex to implement in practical language models.
Innovation Solution
A speech processing apparatus and method that extracts recognition feature information and language feature information to create a verification model using a learning process, incorporating conditional random fields, to improve the accuracy of speech recognition results by considering a broader range of linguistic features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only recognition features from speech recognizing unit are used for confidence measure evaluation, then the evaluation process is simple, but the linguistic accuracy and confidence measure evaluation accuracy are insufficient
Solution Approach 1:
The patent merges recognition features (from speech recognizing unit) with language features (from language model) to create a comprehensive verification model. This combination allows the system to evaluate both recognition confidence and linguistic accuracy, resolving the contradiction by integrating multiple feature sources to improve evaluation accuracy without overwhelming complexity
Solution Approach 2:
The verification model serves multiple functions: it evaluates recognition confidence measures, verifies linguistic accuracy, and identifies ungrammatical sentences. By making the model multi-functional, the system achieves high evaluation accuracy across different aspects without requiring separate simple evaluation mechanisms for each function
2Reliability
If comprehensive language model features are integrated into verification, then verification accuracy improves, but the complexity of implementing language model increases
Solution Approach 1:
The patent segments the verification process into distinct components: recognition feature extraction, language feature extraction, and verification model integration. This segmentation allows the complex verification task to be broken down into manageable parts, improving reliability while controlling implementation complexity through modular architecture
Solution Approach 2:
The verification model acts as an intermediary that bridges the speech recognizing unit and the language model. It integrates features from both sources and produces verification results, thereby improving verification accuracy while managing complexity through this intermediate verification layer rather than directly integrating all features into a single complex model
Data Source
AI summary
A speech processing apparatus 101 includes a recognition feature extracting unit 12 that extracts recognition feature information which is a characteristic of a speech recognition result 15 obtained by performing a speech recognition process on an inputted speech from the speech recognition result 15; a language feature extracting unit 11 that extracts language feature information which is a characteristic of a pre-registered language resource 14 from the language resource 14; and a model learning unit 13 that obtains a verification model 16 by a learning process based on the extracted recognition feature information and language feature information.