【摘要】：語音作為人類交流的重要方式之一,越來越多地被深入探究和擴展,而對人類發(fā)聲原理的研究是促使語音技術(shù)進一步發(fā)展的基礎(chǔ),此研究廣泛應(yīng)用于語音識別、語音合成和相關(guān)醫(yī)學(xué)診斷等領(lǐng)域。對發(fā)聲過程進行建模和仿真,對發(fā)聲機理的深入研究有重要的作用。對聲帶振動模型的研究能夠形象地描述聲帶振動特性,通過深入這種研究,有助于對人體發(fā)聲過程的進一步研究和實際應(yīng)用。本文從發(fā)聲的生理過程出發(fā),通過對國內(nèi)外已有的有關(guān)語音發(fā)聲模型的調(diào)研分析,對語音發(fā)聲模型展開了細(xì)致研究和仿真。首先介紹發(fā)音系統(tǒng)的各個子系統(tǒng)的生理結(jié)構(gòu)、功能特性,并具體分析了聲帶振動發(fā)音過程的原理。在此理論之上,對雙質(zhì)體激勵源模型和多管級聯(lián)的聲道模型進行了分析和建模。先根據(jù)聲帶的分層結(jié)構(gòu),引出雙質(zhì)體激勵源模型,針對發(fā)音過程中的聲道特性,給出聲管級聯(lián)的聲道模型。然后利用電聲類比的方法為聲道傳輸特性建模。對這兩個模型,使用Java編程進行了計算機編程仿真。使用差分方程來對模型進行了離散化,然后通過仿真程序合成出若干濁音信號,給出合成濁音信號的頻譜圖、時域波形圖,并以表格的形式列出了頻域共振峰,在表格中與典型值進行了對比,從主觀的聽覺、客觀的共振峰數(shù)值驗證了仿真的有效性。另一方面,通過仿真程序進行多個發(fā)音參數(shù)的調(diào)整(包括聲帶緊張度、肺部氣壓、聲門初始面積、聲道長度等),實現(xiàn)了對不同發(fā)音特點的語音信號的有效仿真合成。
[Abstract]:As one of the important ways of human communication, speech is more and more deeply explored and expanded. The research on the principle of human phonation is the basis of the further development of speech technology, which is widely used in speech recognition. Speech synthesis and related medical diagnostics. Modeling and simulation of vocal process plays an important role in the study of sound mechanism. The study of vocal cord vibration model can describe the vibration characteristics of vocal cord vividly. The further study will be helpful to the further study and practical application of human vocal process. Based on the physiological process of phonation and through the investigation and analysis of the existing phonetic models at home and abroad, this paper makes a detailed study and simulation of the phonetic models. This paper first introduces the physiological structure and functional characteristics of each subsystem of the pronunciation system, and analyzes the principle of the vocal cord vibration pronunciation process in detail. Based on this theory, the dual body excitation source model and the multi-tube cascade sound channel model are analyzed and modeled. According to the layered structure of the vocal cord, the dual body excitation source model is derived, and the sound channel model of the sound tube cascade is given according to the characteristics of the vocal channel in the process of pronunciation. Then the acoustic analogy is used to model the transmission characteristics of the channel. For these two models, the computer programming simulation is carried out by using Java programming. The model is discretized by difference equation, and then some turbidimetric signals are synthesized by simulation program. The spectrum diagram and time domain waveform of the synthesized turbidimetric signals are given, and the resonance peaks in frequency domain are listed in the form of tables. Compared with the typical values in the table, the validity of the simulation is verified by the subjective auditory and objective resonance peak values. On the other hand, the speech signal with different pronunciation characteristics is effectively synthesized by adjusting several phonetic parameters (including vocal cord tension, lung pressure, glottis initial area, vocal channel length, etc.) through the simulation program.
【學(xué)位授予單位】：青島大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TN912.3

【參考文獻】

相關(guān)期刊論文 前5條

1 康曉莉;曹輝;;聲帶雙彈簧對稱振動模型研究[J];陜西師范大學(xué)學(xué)報(自然科學(xué)版);2011年06期

2 蔣景英,虞啟璉,邱慶軍,徐可欣;聲帶振動的二質(zhì)量塊-有限元模型[J];生物醫(yī)學(xué)工程學(xué)雜志;2005年02期

3 賈雷鳴,李迎新,張偉,黃勇旺;聲帶振動波形分析方法的研究[J];天津醫(yī)科大學(xué)學(xué)報;2001年03期

4 鄭義，蔣剛毅，張禮和，，蔡金海;基于聲帶振動模型和聲門波的嘶音研究[J];聲學(xué)學(xué)報;1996年06期

5 張禮和,程啟明;嘶音的三質(zhì)量塊聲帶模型分析方法[J];中國生物醫(yī)學(xué)工程學(xué)報;1991年04期

相關(guān)博士學(xué)位論文 前2條

1 邵駿;聲帶振動動態(tài)特征的定量研究[D];復(fù)旦大學(xué);2009年

2 吳永忠;基于物理模型的聲音合成技術(shù)理論研究[D];合肥工業(yè)大學(xué);2003年

相關(guān)碩士學(xué)位論文 前1條

1 潘欣裕;漢語耳語音特征分析與應(yīng)用研究[D];蘇州大學(xué);2007年

本文編號：2296522