Contextual Attribute Provision for Natural Speech Synthesis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional speech synthesis systems fail to provide attribute information contextually across multiple sentences, leading to unnatural emotional or speech style transitions, as they typically associate attribute information with specific character strings rather than considering the broader context or relationships between sentences.

Innovation Solution

An information provision system that divides linguistic expressions into predetermined units, determines attribute information using a dictionary, and provides it contextually to linguistic units based on their connecting relationships, ensuring appropriate emotional or speech style representation across sentences.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If attribute information is provided based on the entire passage or presence of character string expressions, then the system is simple to operate, but the emotional or speech style representation becomes unnatural and lacks context

Engineering Contradiction:

Improvenaturalness of emotional expressionVSAvoidcomplexity of attribute information processing

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The passage is divided into multiple linguistic units (sentences, phrases, or clauses) rather than treating it as a single block. Each linguistic unit is independently analyzed for attribute information while considering its relationship with other units, enabling contextually appropriate emotional expression without overwhelming complexity

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different linguistic units within the same passage are assigned different attribute information based on their specific context and connecting relationships. This allows each unit to have its own emotional or speech style characteristics rather than applying a uniform attribute to the entire passage

Inventive Principle:

Principle #3Local quality

2Reliability

If attribute information is provided only to segments with specific character string expressions, then the processing is simple, but the emotional transition between segments becomes abrupt and unnatural

Engineering Contradiction:

Improvesmoothness of emotional transitionVSAvoidprocessing efficiency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system performs preliminary analysis of connecting relationships between linguistic units before final attribute information assignment. By pre-identifying how units are related (sequential, parallel, causal, etc.), the system can smoothly transition emotions across units without reprocessing during output generation

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The connecting relationship analysis acts as an intermediary mechanism that bridges linguistic units with explicit attribute expressions and units without them. This intermediary enables indirect propagation of attribute information through contextual relationships, ensuring smooth emotional transitions while maintaining processing efficiency

Inventive Principle:

Principle #24Intermediary (Mediator)

3Loss of information

If the system analyzes connecting relationships between linguistic units to provide contextual attribute information, then the expressiveness improves, but the computational complexity increases

Engineering Contradiction:

Improvecontextual information retentionVSAvoidcomplexity of relationship analysis

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The connecting relationship analysis is segmented into predefined relationship types (sequential, parallel, causal, adversative, etc.), allowing the system to handle complex contextual relationships through a structured classification framework rather than attempting to analyze all possible relationships simultaneously

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The attribute information provision mechanism is designed to handle multiple types of linguistic units and connecting relationships through a unified framework. The same basic process of relationship analysis and attribute propagation applies regardless of the specific unit types or relationship kinds, reducing overall system complexity

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS8340956B2Information provision system, information provision method, information provision program, and information provision program recording medium

Publication Date: 2012.12.25 CLOUD BYTE LLC

AI summary

An information provision system capable of providing attribute information to a contextually appropriate portion as well as a linguistic unit including a specific character string expression is provided. The information provision system includes analysis means 21 for dividing a linguistic expression into predetermined linguistic units, a dictionary 31 for extracting vocabularies each of which determines attribute information and selecting a linguistic unit to which the attribute information is to be provided, attribute information determination means (first attribute information determination means) 22 for extracting a predetermined vocabulary from the linguistic unit and determining the attribute information using the dictionary, and attribute information provision linguistic unit selection means (second attribute information determination means) 23 for determining the attribute information on an adjacent linguistic unit, based on the attribute information determined by the first attribute information determination means 22 and a connecting relationship between the respective linguistic units.