Multi-Language Voice Recognition Using Context-Based Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current voice recognition methods for multiple languages are inefficient, requiring excessive calculations and unable to operate in real-time, especially when dealing with mixed languages, leading to difficulties in identifying non-primary languages and processing voices effectively.

Innovation Solution

An apparatus and method that segment input voices into phonemes, use a phoneme database for matching, and determine the type of non-primary language based on context information such as location, dialogue history, and user feedback, allowing for real-time recognition by applying specific language algorithms to primary and non-primary language components.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If mapping is performed on all languages to recognize voice involving multiple languages, then the recognition capability is improved, but the amount of calculations rapidly increases

Engineering Contradiction:
Improverecognition capabilityVSAvoidamount of calculations
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments the voice recognition process into two distinct phases: first identifying the language type using context information (location, dialogue history, user feedback), then performing phoneme matching only for the identified language type. This segmentation avoids the need to perform comprehensive mapping on all languages simultaneously, thereby reducing computational complexity while maintaining multi-language recognition capability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary language type identification before executing the main voice recognition task. By using context information to determine which language type is being spoken first, the system prepares the appropriate phoneme database and recognition parameters in advance, avoiding unnecessary calculations for other language types and enabling real-time processing.

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If mapping is performed on all languages to recognize voice involving multiple languages, then the recognition capability is improved, but real-time processing cannot be achieved

Engineering Contradiction:
Improverecognition capabilityVSAvoidreal-time processing
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent divides the recognition system into a language type identification module and a phoneme recognition module. The identification module uses lightweight context information processing to quickly determine language type, while the phoneme recognition module processes only the relevant language data. This segmentation enables real-time processing by avoiding the computational burden of simultaneous multi-language mapping.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs preliminary language type identification using context information before initiating the main recognition process. This preliminary action allows the system to configure the appropriate phoneme database and recognition parameters in advance, enabling real-time voice recognition without the delay of comprehensive language mapping during the recognition phase.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If phoneme matching is performed for all languages, then recognition accuracy is improved, but computational load increases

Engineering Contradiction:
Improverecognition accuracyVSAvoidcomputational load
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent applies local quality by loading and performing phoneme matching only with the phoneme database corresponding to the identified language type, rather than maintaining and processing all language databases simultaneously. This approach ensures high recognition accuracy for the actual language being spoken while significantly reducing the computational load and memory requirements.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system performs phoneme matching only for the partial set of phonemes relevant to the identified language type, rather than performing exhaustive matching across all languages. This partial action maintains recognition accuracy for the target language while avoiding the excessive computational burden of processing all possible language phonemes.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2821991B1Apparatus and method for recognizing voice and text
Publication Date: 2018.11.14 SAMSUNG ELECTRONICS CO LTD
  • EP2821991B1 patent drawingFigure 1
  • EP2821991B1 patent drawingFigure 2A~2D
  • EP2821991B1 patent drawingFigure 3A

AI summary

A method for recognizing a voice includes receiving, as an input, a voice involving multiple languages, recognizing a first voice of the voice by using a voice recognition algorithm matched to a preset primary language, identifying the preset primary language and a non-primary language different from the preset primary language, which are included in the multiple languages, determining a type of the non-primary language based on context information, recognizing a second voice of the voice in the non-primary language by applying a voice recognition algorithm, which is matched to the non-primary language of the determined type, to the second voice, and outputting a result of recognizing the voice which is based on a result of recognizing the first voice and a result of recognizing the second voice.