The Mongolian Vowel Acoustic Model Based on the Clustering Algorithm

Area of Science:

• Computational Linguistics
• Speech Processing
• Machine Learning

Background:

• Challenges in accurately modeling Mongolian vowels for speech recognition systems.
• Need for robust acoustic models to handle linguistic and acoustic variations in Mongolian.

Purpose of the Study:

• To propose a vowel acoustic model using clustering algorithms and speech recognition technology for the Mongolian language.
• To enhance the accuracy and technical support for Mongolian speech recognition systems.

Main Methods:

• Development of a Mongolian vowel acoustic database with increased sample size.
• Implementation of classification and context modeling techniques.
• Construction of language vowel and Mongolian speech recognition systems.

Main Results:

• The proposed model achieved a word recognition accuracy of 86% and sentence recognition accuracy of 45%.
• Experimental results demonstrated an improvement of over 2% compared to previous methods.
• Sparse tritones analysis confirmed the model's effectiveness.

Conclusions:

• The developed vowel acoustic model provides significant improvements in Mongolian speech recognition accuracy.
• The model offers valuable technical support for Mongolian language learning and pronunciation tools.
• Further research can expand upon these findings for broader applications.