Speech Engine Phonetic Alphabet Conversion via Reference Representation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing speech processing engines face limitations in efficiently setting up and using different phonetic alphabets, as associations between written words and phonetic representations are not easily transferable across engines, restricting the selection of the best-suited engine and reducing performance.

Innovation Solution

A computer-implemented method that accesses a table of conversion between multiple phonetic alphabets and a reference representation, allowing users to convert associations between a first phonetic alphabet and a second, enabling seamless setup of speech processing engines by generating equivalent associations, and integrating these into dictionaries or training databases.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If users provide associations between written words and phonetic representations for each speech processing engine separately, then each engine can be optimized for its specific phonetic alphabet, but the setup time and effort increase significantly

Engineering Contradiction:
Improveengine performanceVSAvoidsetup time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent creates a reference phonetic representation that can be copied and converted to different phonetic alphabets. Instead of manually creating associations for each engine, the system copies the reference representation and converts it to the target phonetic alphabet automatically, dramatically reducing setup time while maintaining engine-specific optimization.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent performs preliminary action by creating a reference phonetic representation in advance that contains the essential phonetic information. This reference can then be quickly converted to any target phonetic alphabet when needed, avoiding the need to create associations from scratch for each engine setup.

Inventive Principle:
Principle #10Preliminary action

2Adaptability or versatility

If different speech processing engines use different phonetic alphabets, then each engine can be optimized for its specific requirements, but associations between written words and phonetic representations cannot be transferred between engines

Engineering Contradiction:
Improveengine selection flexibilityVSAvoidassociation transferability
Core Design Contradiction:
Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent introduces a reference phonetic representation as an intermediary between different phonetic alphabets. This reference representation serves as a universal mediator that can be converted to any target phonetic alphabet, enabling associations to be transferred between engines with different phonetic requirements while maintaining each engine's specific optimization.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The reference phonetic representation is designed to be universal and multi-functional, serving as a common basis that can be adapted to any phonetic alphabet. This universal reference enables the same association data to be used across multiple engines with different phonetic requirements, improving both transferability and flexibility.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Manufacturing precision

If users are required to learn and use the specific phonetic alphabet of a speech processing engine, then accurate associations can be created, but the ease of use decreases for users familiar with different phonetic alphabets

Engineering Contradiction:
Improveassociation accuracyVSAvoiduser familiarity
Core Design Contradiction:
Manufacturing precisionVSEase of operation

Solution Approach 1:

The reference phonetic representation acts as an intermediary that users can work with regardless of their familiarity with specific phonetic alphabets. Users create associations using the universal reference, and the system automatically handles the conversion to the engine-specific phonetic alphabet, maintaining accuracy while improving ease of use for users with diverse phonetic knowledge backgrounds.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4261822A1Setting up of speech processing engines
Publication Date: 2023.10.18 VIVOKA
  • EP4261822A1 patent drawingFigure 1
  • EP4261822A1 patent drawingFigure 2
  • EP4261822A1 patent drawingFigure 3~4

AI summary

The invention relates to a computer-implemented method comprising the steps of: accessing a table of conversion between a plurality of phonetic alphabets and a reference representation; receiving from a user a first association between a word and a series of phonemes in a first phonetic alphabet belonging to said plurality of phonetic alphabets; obtaining a selection of a speech processing engine using a second phonetic alphabet belonging to said plurality of phonetic alphabets and different from the first phonetic alphabet; converting said series of phonemes in said first phonetic alphabet into a series of elements of said reference representation using said table; converting said series of elements said reference representation into a series of phonemes in said second phonetic alphabet using said table; generating a second association between said word and said series of phonemes in said second alphabet; setting up said speech processing engine using said second association.