Natural Language Command Extraction for Wake-Word-Free Device Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing devices struggle to accurately extract and execute commands from natural user utterances without requiring a wake-up word or specific command recognition, making it difficult to control devices effectively in everyday conversations.

Innovation Solution

A method and computing apparatus utilize a language model to identify and extract target text from user utterances, distinguishing between target commands and non-target text to control devices accordingly.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a wake-up word or specific command recognition is required, then device control reliability is improved, but user interaction complexity and operation time increase

Engineering Contradiction:
Improvedevice control reliabilityVSAvoiduser interaction complexity
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The patent extracts the command from within the natural language utterance using NLU technology, separating the actionable instruction from the conversational context. This allows the system to identify and execute commands without requiring dedicated wake-up words or structured command formats, thus maintaining reliability while simplifying user interaction.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces an NLU module as an intermediary between the user's natural language input and the device control system. This mediator translates casual conversation into executable commands, bridging the gap between natural speech and structured control instructions without requiring users to learn specific trigger phrases.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Ease of operation

If natural language utterances are processed without wake-up words, then ease of operation is improved, but command extraction accuracy deteriorates

Engineering Contradiction:
Improveuser interaction simplicityVSAvoidcommand extraction accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent performs preliminary processing of the utterance through speech-to-text conversion and NLU analysis before command execution. By pre-processing the natural language input through trained models that understand context and intent, the system maintains high command extraction accuracy even without wake-up words, as the NLU module is prepared to identify commands within the full conversational context.

Inventive Principle:
Principle #10Preliminary action

3Device complexity

If the entire utterance is processed as a command, then command extraction simplicity is improved, but loss of information increases due to inclusion of non-command text

Engineering Contradiction:
Improvecommand extraction process complexityVSAvoidnon-command text filtering
Core Design Contradiction:
Device complexityVSLoss of information

Solution Approach 1:

The patent segments the utterance into distinct components using NLU technology, separating the actionable command from surrounding conversational text. The system divides the input into intent, entities, and context elements, allowing precise extraction of only the relevant command portions while discarding unnecessary conversational filler, thus reducing information loss without overly complicating the process.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4682869A1Method for controlling device on basis of command extracted from user utterance and computing apparatus for performing same
Publication Date: 2026.01.21 SAMSUNG ELECTRONICS CO LTD
  • EP4682869A1 patent drawingFigure 1
  • EP4682869A1 patent drawingFigure 2
  • EP4682869A1 patent drawingFigure 3

AI summary

According to the present disclosure, a method includes obtaining an utterance of a user, determining whether a target text requesting a device to perform a function is included in the utterance by using a language model, and controlling the device based on the target text based on determining that the target text is included, wherein the language model may be a model trained to extract text related to a request from successive sentences.