Utterance Cache Routing for Low-Latency Voice Intent Execution

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing speech recognition systems struggle to accurately identify user intents and provide appropriate content services due to limitations in processing user utterances efficiently, leading to inefficiencies and increased latency.

Innovation Solution

An electronic device equipped with an utterance cache and processor that identifies cached utterances, executes operations locally when possible, and transfers operations to external devices when necessary, while managing cache size and latency information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If all user utterances are processed by the speech recognition system, then comprehensive service coverage is achieved, but processing time and latency increase

Engineering Contradiction:

Improveservice coverageVSAvoidprocessing latency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system pre-processes and caches utterances that have been previously processed by the speech recognition system. When a new utterance is received, the system first checks the cache to see if an identical or similar utterance exists, thereby avoiding redundant processing and reducing latency while maintaining comprehensive service coverage.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements a hybrid processing approach where different types of utterances are handled differently: cached utterances are processed locally with immediate retrieval, while uncached utterances are processed by the full speech recognition system. This local quality differentiation optimizes the balance between service coverage and processing latency.

Inventive Principle:

Principle #3Local quality

2Productivity

If the utterance cache stores more cached utterances, then processing speed for repeated utterances improves, but memory consumption increases

Engineering Contradiction:

Improveprocessing speedVSAvoidmemory consumption

Core Design Contradiction:

ProductivityVSQuantity of substance

Solution Approach 1:

The system extracts only the essential elements needed for cache matching (the utterance text itself) while storing additional metadata (cache status, processing results) separately. This extraction approach minimizes the memory footprint of the cache while maintaining the ability to quickly retrieve and process repeated utterances.

Inventive Principle:

Principle #2Taking out (Extraction)

3Loss of time

If the system checks for cached utterances before processing, then latency for repeated utterances is reduced, but system complexity increases

Engineering Contradiction:

Improveutterance processing latencyVSAvoidsystem architecture complexity

Core Design Contradiction:

Loss of timeVSDevice complexity

Solution Approach 1:

The system segments the utterance processing workflow into distinct modules: a cache checking module that performs quick lookups, a speech recognition module for full processing, and a cache management module for updating stored utterances. This segmentation allows each module to be optimized independently and simplifies the overall system architecture by creating clear separation of concerns.

Inventive Principle:

Principle #1Segmentation

4Reliability

If the system transmits uncached utterances to external devices, then processing capability is enhanced, but communication overhead increases

Engineering Contradiction:

Improveprocessing capabilityVSAvoidcommunication overhead

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The system converts the potential harm of increased communication overhead into a benefit by implementing intelligent cache management. The cache stores frequently occurring utterances locally, eliminating the need for repeated transmissions to external devices. This approach transforms what would be continuous energy-consuming communications into selective, optimized transmissions only when necessary, thereby reducing overall communication overhead while maintaining enhanced processing capability.

Inventive Principle:

Principle #22Blessing in disguise (Convert harm into benefit)

Data Source

PatentUS12586587B2Method for analyzing user utterance based on utterance cache and electronic device supporting the same

Publication Date: 2026.03.24 SAMSUNG ELECTRONICS CO LTD

AI summary

An electronic device includes: a communication interface; a memory including an utterance cache; and at least one processor operatively connected to the communication interface and the memory, wherein the at least one processor is configured to: identify whether an input utterance corresponds to one or more cached utterances stored in the utterance cache; based on identifying that the input utterance corresponds to the one or more cached utterances, transmit the input utterance to an external electronic device through the communication interface; identify whether an operation corresponding to the input utterance is executable by the electronic device; based on the operation being identified to be executable by the electronic device, execute the operation; and based on the operation being identified not to be executable by the electronic device, transmit the input utterance to the external electronic device through the communication interface and add the input utterance to the utterance cache.