Reading Assessment via Audio-to-Text Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Intelligent devices face challenges in accurately determining the reading quality of users, particularly due to errors in speech recognition, which can lead to incorrect assessment of pronunciation accuracy, especially in cases of younger readers or those with stuttering, nationality, or place of origin, resulting in inefficient learning experiences.
Innovation Solution
A method and apparatus that output a to-be-read audio, perform speech recognition on the user's read audio, calculate the similarity between the read and to-be-read audio based on character strings, and determine a similarity interval to provide a reading evaluation, allowing for improved learning efficiency and interest without sending the read audio file to a backend server.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If speech recognition is used to determine reading quality, then reading assessment can be automated, but errors in speech recognition lead to incorrect assessment of pronunciation accuracy
Solution Approach 1:
The patent introduces character strings as an intermediary between speech recognition results and pronunciation assessment. Instead of directly using speech recognition output to determine pronunciation accuracy, the system converts both the standard audio and user's audio into character strings, then compares these strings to assess pronunciation. This intermediary step filters out speech recognition errors while preserving genuine pronunciation differences.
Solution Approach 2:
The patent creates a copy of the audio content in a different form (character strings) for comparison purposes. By converting audio signals into text representations and then comparing these text copies, the system avoids the pitfalls of direct audio comparison or reliance on potentially erroneous speech recognition results, achieving more accurate pronunciation assessment.
2Measurement precision
If character string comparison is used to calculate similarity, then speech recognition errors are mitigated, but the method becomes more complex for handling different languages and scripts
Solution Approach 1:
The patent implements a universal character string comparison method that can handle multiple languages and scripts through a unified approach. The system determines the language type (Chinese, English, or other) and applies appropriate comparison strategies for each, but the overall framework remains consistent. This multi-functional capability allows the same basic methodology to work across different linguistic contexts without requiring entirely separate systems for each language.
Solution Approach 2:
The patent adjusts comparison parameters based on language type. For Chinese characters, it compares character-by-character; for English and other languages, it may use different comparison criteria. By dynamically changing the comparison parameters according to the language being assessed, the system maintains high accuracy across diverse languages while keeping the underlying methodology consistent.
3Measurement precision
If phonetic character strings are used for cross-language comparison, then pronunciation assessment becomes more accurate for non-English languages, but the processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary language identification and determines the appropriate comparison method before conducting the actual pronunciation assessment. By pre-determining whether to use phonetic character strings or direct character comparison based on language type, the system avoids unnecessary computational overhead while maintaining accuracy for cross-language assessment.
Solution Approach 2:
The patent segments the pronunciation assessment process into distinct stages: language identification, appropriate comparison method selection, and execution. This segmentation allows the system to apply computationally intensive phonetic character string comparison only when necessary (for cross-language scenarios), while using simpler methods for same-language comparisons, thus reducing overall processing time.
Data Source
AI summary
Embodiments of the present disclosure relate to a method and apparatus for outputting information. The method includes: outputting a to-be-read audio in response to receiving a reading instruction from a user; acquiring an actually read audio obtained by reading the to-be-read audio by the user; performing speech recognition on the actually read audio to obtain a recognition result; calculating a similarity between the actually read audio and the to-be-read audio based on a character string corresponding to the recognition result and a character string corresponding to the to-be-read audio; determining, from a predetermined set of similarity intervals, a similarity interval to which the calculated similarity belongs; and outputting a reading evaluation corresponding to the determined similarity interval. The embodiment may help a reader to improve the learning efficiency and learning interest, thereby improving the rate of a user using a device.