Vocabulary Generation Using Unknown Code Words for Speech Recognition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional speech recognition systems are limited by a predetermined vocabulary and syntax, making it inconvenient for users who are unsure of the commands, as they cannot use unknown or forgotten vocabulary, thereby restricting the system's flexibility and user convenience.

Innovation Solution

A vocabulary generating apparatus and method that includes a tokenize unit, an unknown code word (UCW) database, and a vocabulary generating unit, which tokenizes original system vocabulary into meaningful units, combines them with UCWs to create new vocabulary, allowing users to use UCWs for undetermined parts, thereby increasing system flexibility.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the speech recognition system uses a predetermined vocabulary and syntax, then the system can accurately recognize commands, but the user cannot use unknown or forgotten vocabulary, reducing flexibility and convenience

Engineering Contradiction:

ImproveflexibilityVSAvoidconvenience

Core Design Contradiction:

Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent segments the vocabulary into meaningful units (MUs) through tokenization, allowing users to combine these units with unknown code words (UCWs) to form new commands. This segmentation enables flexible command construction without requiring users to memorize the complete predetermined vocabulary, thereby improving both flexibility and ease of operation.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces unknown code words (UCWs) as intermediary elements that can replace or combine with meaningful units. These UCWs act as placeholders or wildcards that allow users to express commands even when they are unsure of the exact vocabulary, thus enhancing convenience while maintaining system accuracy through the mediation of UCW database and vocabulary generating unit.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If the user memorizes the full vocabulary, then the system can accurately recognize commands, but the user experience is degraded due to the need to remember many commands

Engineering Contradiction:

Improvecommand recognition accuracyVSAvoiduser convenience

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The patent applies partial action by requiring users to memorize only a subset of meaningful units (MUs) rather than the complete vocabulary. Users can combine these partial MUs with UCWs to form commands, reducing the memorization burden while maintaining sufficient command recognition accuracy through the systematic combination of partial elements.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent creates universality by making UCWs and meaningful units multi-functional components that can be combined in various ways to generate multiple commands. A single MU can be paired with different UCWs to create different commands, and UCWs can function as placeholders for multiple possible words, thereby reducing the total number of items users need to memorize while maintaining comprehensive command recognition capability.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS8417510B2Vocabulary generating apparatus and method, speech recognition system using the same

Publication Date: 2013.04.09 DELTA ELECTRONICS INC(CN)

AI summary

A vocabulary generating method and apparatus and a speech recognition system using the same are disclosed. In the vocabulary generating method, a new system vocabulary can be generated to increase the flexibility of the speech recognition system, so a user, when unsure of a system command, can use a specially defined “unknown code word” (UCW) for the undetermined part in the command.