Speech Recognition Follow-Up Processing After Session Termination

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current speech recognition technologies in electronic devices often fail to process follow-up utterances after a session has terminated, leading to user inconvenience as these utterances are recognized as new initial utterances rather than follow-ups, disrupting the continuity of service.

Innovation Solution

The method and device extend the support for follow-up utterances after session termination by reprocessing subsequent utterances based on the response results from previous utterances, ensuring that they are recognized and processed correctly even after the initial session has ended, using a designated capsule or domain to maintain consistency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the session is terminated after processing a first utterance, then the system can release resources and maintain clean state, but follow-up utterances cannot be processed as continuations and are treated as new initial utterances

Engineering Contradiction:

Improvefollow-up utterance processing accuracyVSAvoidsession management complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary action by storing capsule information and response results before session termination. This allows the system to quickly retrieve and use this stored information when processing follow-up utterances after session termination, avoiding the need for complex real-time session maintenance while ensuring accurate follow-up processing.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary mechanism (capsule information storage and retrieval system) that bridges the gap between session termination and follow-up utterance processing. This intermediary allows the system to maintain the ability to process follow-ups without actually maintaining an active session, thus resolving the contradiction between resource release and processing capability.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Ease of operation

If the system maintains session continuity to process follow-up utterances, then user convenience is improved, but communication resources are consumed and session termination cannot be executed

Engineering Contradiction:

Improveuser convenienceVSAvoidcommunication resource consumption

Core Design Contradiction:

Ease of operationVSUse of energy by moving object

Solution Approach 1:

The system extracts the essential information needed for follow-up processing (capsule information and response results) from the session and stores it separately. This allows the system to terminate the session and release communication resources while still maintaining the capability to process follow-up utterances using the extracted information, thus improving user convenience without continuous resource consumption.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system treats sessions as disposable short-living objects that are terminated after processing initial utterances. Instead of maintaining expensive long-lived sessions, the system uses cheap stored capsule information to handle follow-ups, achieving user convenience with minimal resource investment.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

3Loss of energy

If follow-up utterances are processed as new initial utterances after session termination, then resource consumption is reduced, but processing accuracy and user experience deteriorate

Engineering Contradiction:

Improveresource consumptionVSAvoidutterance recognition accuracy

Core Design Contradiction:

Loss of energyVSMeasurement precision

Solution Approach 1:

The system creates a copy of the essential session context (capsule information) that can be reused for processing follow-up utterances. This copy allows the system to process follow-ups with high accuracy without maintaining the original session, thus reducing resource consumption while preserving processing precision.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS11756575B2Electronic device and method for speech recognition processing of electronic device

Publication Date: 2023.09.12 SAMSUNG ELECTRONICS CO LTD

AI summary

An electronic device and method are disclosed. The device includes a memory and speech recognition circuitry and/or a processor, which implements the method, including: receiving a first utterance, and processing the first utterance to initiate a session and generate a first response result, after the session related to the first utterance is terminated, receiving a second utterance, processing the second utterance to generate a second response result, based on the second response result, determining whether to execute follow-up utterance processing on the second utterance as if the session were active, based on determining to execute the follow-up utterance processing, reprocessing the second utterance based at least in part on the first response result related to the first utterance to generate a third response result, and outputting the third response result.