語音辨識與人機互動

臺北科技大學 114-2 學期課程，課號 353672。

課程大綱

本課程將引導學生了解語音辨識在人機互動應用上所牽涉的各種技術，包括語音訊號分析、音素/音節辨識、字詞辨識、連續語音辨識、語意辨識、語者辨識、語言辨識、語音合成等，其中也介紹動態時間伸縮與音高比對方法、隱藏式馬可夫模型、類神經網路與深度學習、語者與環境調適、語言模型等方法理論，配合「問題導向學習」之教學模式，進行分組討論、實作練習、企業參訪、與成果觀摩及報告，激發學生之創造與創新能力。 This course guides students to learn the techniques involved in the human-machine interaction based on speech recognition, including speech signal analysis, phoneme/syllable recognition, word recognition, continuous speech recognition, semantic recognition, speaker recognition, language recognition, speech synthesis. The theoretical fundamentals on dynamic time warping, hidden Markov model, neural network/deep learning, speaker and environment adaptation, language modeling, etc. are also introduced. In addition, the course is designed with the Problem-Based Learning (PBL) method, which includes group discussion and collaboration on working projects, corporate visit, and project presentation, etc., so that students' capabilities of creativity and innovation can be developed or enhanced.

課程進度

第1週 導論 第2-4週 音訊號分析、音素/音節辨識 第5-8週 字詞辨識、連續語音辨識、語意辨識、語者辨識、語言辨識、語音合成 第10-11週 動態時間伸縮與音高比對方法、隱藏式馬可夫模型 第12-14週 類神經網路與深度學習 第15-17週 語者與環境調適、語言模型

評量方式

Team work 40%, Final Exam 30%, Homework 30%

教材

【遵守智慧財產權觀念，請使用正版教科書，不得使用非法影印教科書】 使用外文原文書：是 教科書: Language and Speech Recognition for Human-Computer Interaction, by Amitoj Singh,ISBN：9781799813897 教材會放在"北科i學園PLUS" 公告(如考試及作業繳交公告)會放在"北科i學園PLUS"公告欄