Functional Model of Acquisition of Spoken Language Using Categorical Algorithm and Hidden Markov Model

2015 Fiscal Year Final Research Report

• Project/Area Number: 25330201
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Perceptual information processing
• Research Institution: University of the Ryukyus
• Principal Investigator: Takara Tomio 琉球大学, 工学部, 教授 (70163326)
• Co-Investigator(Kenkyū-buntansha): IHA Yasushi 国立沖縄高等専門学校, 教授 (60390564)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 隠れマルコフモデル / クラスタ化 / 言語獲得 / モデル / 音声模倣

Outline of Final Research Achievements

We constructed a computer model simulating a baby which can acquire spoken language by itself. Acquisition of vowels is modeled as a clustering of clear parts of speech. Acquisition of words is modeled as a clustering of the hidden Markov model. Consonants are modeled as the different sound part of two words which has the same vowel sequences. Self-training of speech by a baby is modeled by articulatory parameters and the genetic algorithm. It was shown that we can construct a human acquisition model of spoken language if we use the categorical algorithm and the hidden Markov model.

Free Research Field

知覚情報処理・知能ロボティクス