Dynamic Voice Dictionary Management for Accuracy and Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In typical voice recognition processes, there is a trade-off between reducing computation amount and improving recognition accuracy, which can lead to impaired user experience due to discrepancies between limited utterance expression sets and actual user usage patterns.

Innovation Solution

An information processing apparatus and method that dynamically updates utterance expressions in a dictionary based on user usage performance, deleting low-usage expressions and transferring them to a secondary dictionary, while adding high-priority expressions from a secondary dictionary to the primary dictionary as needed.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a large number of utterance expressions are registered in the dictionary to improve voice recognition accuracy, then recognition accuracy is improved, but computation amount increases

Engineering Contradiction:
Improvevoice recognition accuracyVSAvoidcomputation amount
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent divides the dictionary into a first dictionary for frequently used utterance expressions and a second dictionary for less frequently used expressions. This segmentation allows the system to prioritize processing of high-frequency expressions while maintaining access to lower-frequency ones, thereby improving recognition accuracy for common phrases without proportionally increasing computation for all expressions.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different handling strategies to different parts of the dictionary based on usage frequency. High-frequency expressions in the first dictionary receive priority processing and faster access, while low-frequency expressions in the second dictionary are processed with lower priority. This local differentiation optimizes computation by allocating resources according to actual usage patterns rather than treating all expressions equally.

Inventive Principle:
Principle #3Local quality

2Quantity of substance

If a limited set of utterance expressions is used to reduce computation amount, then computation amount is reduced, but voice recognition accuracy deteriorates

Engineering Contradiction:
Improvecomputation amountVSAvoidvoice recognition accuracy
Core Design Contradiction:
Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent implements dynamic dictionary management where the system adapts to user usage patterns over time. Frequently used utterance expressions are automatically identified and moved to the first dictionary, while less frequently used expressions are moved to the second dictionary. This dynamic adjustment ensures that the computation amount remains optimized while recognition accuracy improves for actually used phrases, preventing deterioration despite limited resources.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter of dictionary organization from static to dynamic based on usage frequency. By continuously monitoring and reorganizing expressions according to actual user behavior, the system maintains high recognition accuracy for relevant phrases while keeping the active dictionary size manageable, thus preventing accuracy deterioration that would result from a fixed limited set.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If a static dictionary is used to minimize device complexity, then device complexity is reduced, but adaptability to user usage patterns deteriorates

Engineering Contradiction:
Improvedictionary management complexityVSAvoidadaptability to user usage
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent implements a self-service mechanism where the dictionary automatically reorganizes itself based on usage frequency without requiring complex external management. The system monitors its own usage patterns and autonomously moves expressions between dictionaries, providing adaptability to user behavior while maintaining relatively simple device architecture. This self-organizing capability reduces the need for complex manual dictionary management systems.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent incorporates feedback loops where usage information is continuously collected and used to adjust dictionary organization. This feedback mechanism enables the system to adapt to user patterns dynamically while keeping the control logic relatively simple - the feedback automatically triggers reorganization decisions without requiring complex analytical systems, thus balancing adaptability with manageable complexity.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS12067971B2Information processing apparatus and information processing method
Publication Date: 2024.08.20 SONY GROUP CORP
  • US12067971B2 patent drawing
  • US12067971B2 patent drawing
  • US12067971B2 patent drawing

AI summary

Provided is an information processing apparatus including a first dictionary that has registered a plurality of utterance expressions, a first voice recognition unit that executes a first voice recognition process using the first dictionary, and a dictionary update unit that updates utterance expressions to be registered to the first dictionary based on usage performance of utterance of a user, in which the dictionary update unit deletes an utterance expression having a low usage performance of utterance from the first dictionary or transfers the utterance expression to a second dictionary that has registered a plurality of utterance expressions not registered in the first dictionary.