Progressive Voice UI Semantic Analysis for Real-Time Candidate Display

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In typical voice UI systems, users experience a lack of feedback during speech recognition processing, leading to uncertainty about the status of their voice input until a final response is determined, with no intermediate notification of potential responses.

Innovation Solution

The system performs real-time speech recognition and semantic analysis, displaying candidate response icons or text during the speech process, providing users with immediate feedback on recognized inputs and potential actions through a voice UI.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the system waits until speech recognition is complete before providing feedback, then the accuracy of the final response is improved, but the user experience deteriorates due to lack of intermediate feedback and increased waiting time

Engineering Contradiction:
Improveresponse accuracyVSAvoiduser waiting time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary speech recognition processing during the user's speech and displays candidate responses before the final recognition is complete. This allows the user to see potential responses early, reducing perceived waiting time while the system continues processing to ensure final accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system provides intermediate feedback by displaying candidate responses during speech recognition processing. This feedback mechanism keeps the user informed about the system's processing status and potential interpretations, improving user experience without sacrificing final response accuracy.

Inventive Principle:
Principle #23Feedback

2Ease of operation

If the system displays candidate responses during speech processing, then user feedback is improved through real-time notification, but the system complexity increases due to intermediate processing and display requirements

Engineering Contradiction:
Improveuser feedbackVSAvoidsystem complexity
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The system displays a limited number of high-probability candidate responses rather than all possible interpretations. This partial action approach provides useful feedback to users without requiring the system to process and display every possible response, thereby managing complexity while improving user feedback.

Inventive Principle:
Principle #16Partial or excessive action

3Speed

If the system processes speech recognition in real-time, then the responsiveness is improved, but the processing accuracy may deteriorate due to incomplete utterance analysis

Engineering Contradiction:
Improveprocessing speedVSAvoidsemantic analysis accuracy
Core Design Contradiction:
SpeedVSMeasurement precision

Solution Approach 1:

The system performs preliminary speech recognition to generate candidate responses during the user's speech, then continues processing to refine and confirm the final response. This allows real-time responsiveness through early candidate generation while maintaining accuracy through continued processing.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system dynamically adjusts the display of candidate responses based on processing confidence levels. As recognition progresses and confidence increases, the system can update candidate displays, providing a dynamic balance between speed and accuracy throughout the processing timeline.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3282447B1Progressive utterance analysis for successively displaying early suggestions based on partial semantic parses for voice control. real time progressive semantic utterance analysis for visualization and actions control.
Publication Date: 2020.08.26 SONY GROUP CORP
  • EP3282447B1 patent drawingFigure 1
  • EP3282447B1 patent drawingFigure 2~3
  • EP3282447B1 patent drawingFigure 4

AI summary

[Object] To provide an information processing apparatus, control method, and program capable of notifying a user of a candidate for a response, from the middle of a speech, through a voice UI. [Solution] Provided is an information processing apparatus including: a semantic analysis unit configured to perform semantic analysis on speech text recognized by a speech recognition unit in the middle of a speech; a score calculation unit configured to calculate a score for a response candidate on the basis of a result of the analysis performed by the semantic analysis unit; and a notification control unit configured to perform control to notify of the response candidate, in the middle of the speech, according to the score calculated by the score calculation unit.