Utterance Analysis Apparatus Distinguishing Simple Responses from Statements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for analyzing discussion activity in groups only consider speech duration and not the content of utterances, making it impossible to determine the type of speech, such as simple responses or statements.
Innovation Solution
An information processing apparatus that extracts utterances from voice data, determines utterance periods, and identifies simple responses and statements based on predetermined time periods and keywords, storing the results along with frequent keywords for analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If only speech duration is used for calculation, then the calculation is simple, but the type of utterance cannot be determined
Solution Approach 1:
The speech analysis is segmented into multiple dimensions: duration-based classification (simple response vs. statement) and content-based analysis (keyword extraction). This segmentation allows the system to process different aspects of speech separately, maintaining simplicity in duration analysis while adding content analysis to capture utterance type information.
Solution Approach 2:
The system transitions from one-dimensional duration analysis to two-dimensional analysis by adding content/keyword dimension. Utterances are classified first by duration (temporal dimension) and then enriched with keyword information (semantic dimension), enabling determination of utterance types without significantly increasing computational complexity.
2Loss of information
If speech content is analyzed in addition to duration, then utterance type can be determined, but the calculation becomes more complex
Solution Approach 1:
The analysis is divided into independent modules: duration detection module, keyword extraction module, and classification module. Each module handles a specific aspect, allowing the system to incorporate content analysis without creating a monolithic complex system. The segmentation enables parallel processing and independent optimization of each component.
Solution Approach 2:
The system performs multiple functions using the same processed data: duration-based classification, keyword extraction, and utterance type determination. By making the speech processing system multi-functional, it extracts maximum information from the input data without requiring separate processing pipelines, thereby managing complexity while providing comprehensive analysis.
3Ease of operation
If only speech period is used, then the measurement is simple, but the discussion dynamics cannot be fully understood
Solution Approach 1:
Discussion dynamics are segmented into quantifiable metrics: speech duration, keyword frequency, and utterance type distribution. Each metric captures a different aspect of discussion dynamics, allowing the system to maintain measurement simplicity while providing comprehensive insights into group interactions through multiple independent measurements.
Solution Approach 2:
The system adds semantic dimensions to the temporal measurements. While speech period provides temporal information, keyword extraction and utterance type classification add semantic dimensions, enabling a more complete understanding of discussion dynamics without sacrificing the simplicity of temporal analysis.
Data Source
AI summary
An information processing apparatus as an image forming apparatus includes an utterance period detecting section, a simple response/statement determining section, and an HDD. The utterance period detecting section detects utterance periods of utterances of each person from voice data. The simple response/statement determining section converts the voice data to a text, determines, when the utterance in the detected utterance period falling within a first period contains any predetermined keyword, that the utterance is a simple response, determines the utterance made for a second period longer than the first period to be a statement, and extracts, for each person, a frequent keyword appearing a predetermined number of times or more in the utterances. The HDD stores determination results of the simple response/statement determining section, the utterance periods for the simple responses, and the utterance periods for the statements, together with the frequent keyword.