Lexical Stress Classification via Syllable-Level Feature Vectors

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional language learning systems fail to effectively classify lexical stress in utterances, particularly when multiple syllables in a word are stressed, and often require correct phonetic pronunciation before assessing stress, missing opportunities to correct stress mistakes.

Innovation Solution

A method that generates feature vectors based on prosodic and spectral information to classify lexical stress, allowing for multiple primary stressed syllables and providing syllable-level feedback with more than two stress levels, using Gaussian mixture models to compute posterior probabilities for stress classification.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional systems classify lexical stress only at the word level with a single primary stress, then the system is simple to implement, but it cannot accurately handle cases where multiple syllables are stressed

Engineering Contradiction:
Improvestress classification accuracyVSAvoidclassification system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the word-level stress classification into syllable-level classification. Each syllable is independently analyzed and assigned a stress level, allowing multiple syllables to be identified as primary stressed. This segmentation enables accurate handling of complex stress patterns while maintaining a manageable system architecture through modular processing of individual syllables.

Inventive Principle:
Principle #1Segmentation

2Reliability

If the system requires correct phonetic pronunciation before assessing stress, then phonetic accuracy can be ensured, but stress pronunciation mistakes are missed

Engineering Contradiction:
Improvephonetic pronunciation accuracyVSAvoidstress pronunciation feedback
Core Design Contradiction:
ReliabilityVSLoss of information

Solution Approach 1:

The system performs stress classification independently of phonetic pronunciation verification. By conducting stress analysis as a preliminary or parallel process rather than a sequential step dependent on phonetic accuracy, the system can identify and provide feedback on stress errors even when phonetic pronunciation is imperfect, preventing loss of stress feedback information.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If the system provides detailed syllable-level stress feedback with multiple stress levels, then stress classification precision is improved, but computational complexity increases

Engineering Contradiction:
Improvestress level classification precisionVSAvoidcomputational system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system employs multiple stress level parameters (e.g., primary, secondary, tertiary stress levels) to classify syllables, enabling detailed differentiation of stress patterns. This parametric approach allows precise classification by varying the stress level attribute assigned to each syllable, capturing nuanced pronunciation variations without requiring overly complex system architecture.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS9928832B2Method and apparatus for classifying lexical stress
Publication Date: 2018.03.27 SRI INTERNATIONAL
  • US9928832B2 patent drawing
  • US9928832B2 patent drawing
  • US9928832B2 patent drawing

AI summary

A method for classifying lexical stress in an utterance includes generating a feature vector representing stress characteristics of a syllable occurring in the utterance, wherein the feature vector includes a plurality of features based on prosodic information and spectral information, computing a plurality of scores, wherein each of the plurality of scores is related to a probability of a given class of lexical stress, and classifying the lexical stress of the syllable based on the plurality of scores.