Multilingual Oral Function Estimation Using Universal Speech Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing oral function estimation techniques are inaccurate when applied to speakers of languages different from the specific language used for training, due to language-specific voice feature differences.
Innovation Solution
An estimation device that instructs speakers to repetitively utter syllables containing velar plosives, alveolar fricatives, and alveolar plosives, analyzes voice features, and estimates oral function using a multilingual approach.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If language-specific voice features are used for estimation, then estimation accuracy for that particular language is improved, but the system cannot accurately estimate oral function for speakers of different languages
Solution Approach 1:
The patent applies universality by designing a voice feature extraction system that identifies universal articulatory characteristics common across multiple languages. The estimator is configured to recognize voice features related to tongue movements and oral cavity changes that are consistent across different languages, enabling the same estimation model to work effectively for speakers of various languages without requiring language-specific customization.
2Adaptability or versatility
If universal articulatory features are extracted, then multilingual estimation capability is improved, but language-specific estimation precision may be reduced
Solution Approach 1:
The patent applies local quality by selectively extracting specific voice features that are locally optimal for oral function estimation. Rather than using all possible voice features or features optimized for a particular language, the system identifies and extracts only those features that are both universal across languages and sensitive to oral function changes. This selective feature extraction maintains high estimation accuracy while achieving multilingual capability.
Data Source
AI summary
An estimation device includes: an instructor that instructs a speaker to repetitively utter two syllables containing (i) a sound including a velar plosive in a consonant or a sound including an alveolar fricative in a consonant, and (ii) a sound including an alveolar plosive in a consonant; a voice obtainer that obtains a voice of the speaker; an estimator that analyzes a voice feature of the voice obtained, and estimates an oral function of the speaker based on the voice feature analyzed; and a presenter that presents a condition of the oral function of the speaker which has been estimated.