Terminal Speech Recognition Dictionary Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Speech recognition systems on terminal devices face limitations due to limited computer resources, resulting in fewer recognized words and phrases, and struggles with unique or incorrectly used pronunciations, leading to user inconvenience, especially when communication with a server-side speech recognition engine is unavailable.

Innovation Solution

A terminal device with a dictionary management module that performs morphological analysis and pronunciation selection to register and assign appropriate pronunciations to recognized words and phrases, using both local and server-side recognition results to enhance speech recognition capabilities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a speech recognition engine is installed on the terminal device side to enable offline recognition, then speech recognition can be performed without server communication, but the number of recognizable words and phrases is limited due to limited computer resources

Engineering Contradiction:
Improvespeech recognition availabilityVSAvoidnumber of recognizable words and phrases
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The speech recognition dictionary is segmented into two parts: a basic dictionary stored on the terminal device and an extended dictionary stored on the server. The terminal device uses the basic dictionary for offline recognition, while the server maintains the extended dictionary with additional words and phrases. This segmentation allows the terminal to operate independently with limited resources while still having access to a larger vocabulary when connected to the server.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If the speech recognition dictionary on the terminal device is expanded to recognize more words and phrases, then recognition accuracy improves, but device complexity and resource consumption increase

Engineering Contradiction:
Improverecognition accuracyVSAvoiddictionary size
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

Instead of loading the complete extended dictionary onto the terminal device, the system loads only a subset of dictionary data (basic dictionary) that fits within the terminal's resource constraints. The server handles the remaining dictionary data (extended dictionary), providing partial recognition capability on the terminal while maintaining comprehensive recognition capability on the server.

Inventive Principle:
Principle #16Partial or excessive action

3Productivity

If a pronunciation assignment function is used to assign pronunciations to recognized text, then speech recognition can proceed, but correct pronunciation cannot be assigned to proper nouns with unique pronunciations

Engineering Contradiction:
Improvespeech recognition processingVSAvoidpronunciation accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

A pronunciation dictionary is introduced as an intermediary data structure that stores correct pronunciation information for words and phrases, particularly proper nouns with unique pronunciations. This pronunciation dictionary works in conjunction with the speech recognition dictionary to provide accurate pronunciation assignment. The system includes a pronunciation assignment unit that uses this intermediary dictionary to correctly assign pronunciations even for words with non-standard or unique pronunciations.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3425629B1Speech recognition system, terminal device, and dictionary management method
Publication Date: 2019.11.20 FAURECIA CLARION ELECTRONICS CO LTD
  • EP3425629B1 patent drawingFigure 1
  • EP3425629B1 patent drawingFigure 2
  • EP3425629B1 patent drawingFigure 3

AI summary

To assign an appropriate pronunciation to a word or phrase having a unique pronunciation or a word or phrase having a pronunciation incorrectly used by a user, a terminal device divides a first word or phrase indicated by a first recognition result acquired from a speech recognition server into morphemes and assigns a pronunciation to each of the morphemes, and divides a second word or phrase indicated by a second recognition result acquired from a speech recognition module into morphemes. Further, the terminal device selects, for a morpheme having the same character string as that of any one of the morphemes forming the second word or phrase among the morphemes forming the first word or phrase, a pronunciation of the morpheme indicated by the second recognition result.