Electronic device and method for recognizing voice by same

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing speech recognition systems in environments with noise generated by electronic devices, such as home appliances, face challenges in maintaining high recognition rates and accuracy due to interference from device noise, which varies with operation states.

Innovation Solution

An electronic device and corresponding speech recognition method that dynamically select and adjust speech recognition models and sensitivity based on the operation state, including noise patterns from both the device and surrounding environments, to enhance recognition performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If speech recognition is performed in a noisy environment generated by electronic device operation, then the electronic device can provide speech recognition service during operation, but the speech recognition rate and accuracy deteriorate due to noise interference

Engineering Contradiction:
Improvespeech recognition service availabilityVSAvoidspeech recognition accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary noise measurement and characterization during electronic device operation before speech recognition. The noise pattern is captured and stored in advance, then used to generate a pre-trained speech recognition model that is specifically adapted to the operational noise environment, thereby improving recognition accuracy without sacrificing service availability

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system changes the parameters of the speech recognition model based on the measured noise characteristics. By adjusting the model parameters according to the specific noise pattern generated during device operation, the system optimizes the model's performance for the actual operating conditions, resolving the contradiction between service availability and recognition accuracy

Inventive Principle:
Principle #35Parameter changes

2Device complexity

If the speech recognition model is fixed and not adapted to operation state, then the system complexity is reduced, but the speech recognition accuracy deteriorates under varying noise conditions

Engineering Contradiction:
Improvesystem complexityVSAvoidspeech recognition accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The system implements dynamic adaptation of the speech recognition model based on the electronic device's operation state. The model is adjusted in real-time or near-real-time according to the current noise conditions, allowing the system to maintain high recognition accuracy across varying operational scenarios while managing complexity through state-based model selection

Inventive Principle:
Principle #15Dynamics

3Measurement precision

If noise measurement is performed to improve speech recognition accuracy, then the recognition rate increases, but the measurement and processing time increases

Engineering Contradiction:
Improvespeech recognition accuracyVSAvoidnoise measurement and processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs noise measurement and model generation in advance during device operation, before actual speech recognition is needed. This preliminary preparation allows the speech recognition to proceed quickly using the pre-adapted model, thereby improving accuracy without adding significant delay to the recognition process itself

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3982359B1Electronic device and method for recognizing voice by same
Publication Date: 2024.08.28 SAMSUNG ELECTRONICS CO LTD
  • EP3982359B1 patent drawingFigure 1
  • EP3982359B1 patent drawingFigure 2
  • EP3982359B1 patent drawingFigure 3

AI summary

An electronic device according to an embodiment comprises: a microphone which receives a user's voice; a memory which stores a plurality of voice recognition models; and a processor which selects one voice recognition model from the plurality of voice recognition models stored in the memory, on the basis of the operation state of the electronic device, and recognizes the user's voice received by the microphone, on the basis of the selected voice recognition model.