Prosodic Modification of Chinese Speech Based on Sinusoidal Model

Modification on time scale and pitch scale of Chinese syllable based on sinusoidal model is presented in this paper. Firstly, the short term speech is decomposed into a sum of sinusoidal waves of different magnitudes and phases. Then vocal tract system and excitation are obtained using a homomophic technique. Lastly, the speech with desired time scale and pitch scale is obtained through the change of frequency and phase of excitation while the parameters of vocal tract system are changed accordingly. The results show that the adjustable scale of pitch and time scale is big using this algorithm and it is suitable to be used in analysis and synthesis of Chinese speech.

Emotional Speech Synthesis Based on Prosodic Feature Modification

The synthesis of emotional speech has wide applications in the field of human-computer interaction, medicine, industry and so on. In this work, an emotional speech synthesis system is proposed based on prosodic features modification and Time Domain Pitch Synchronous OverLap Add (TD-PSOLA) waveform concatenative algorithm. The system produces synthesized speech with four types of emotion: angry, happy, sad and bored. The experiment results show that the proposed emotional speech synthesis system achieves a good performance. The produced utterances present clear emotional expression. The subjective test reaches high classification accuracy for different types of synthesized emotional speech utterances.

The Implication of Prosodic Features' Pragmatic Functions-For Teaching Listening and Speaking

Interactive communication is not straightforward but complicated. Prosodic features play an influential role in English communication. They can be used to signal certain pragmatic purposes in real situations for listeners and speakers to have mutual understanding. Identifying the pragmatic functions of prosodic features will facilitate the teaching of listening and speaking. English teachers need to clarify and emphasize the relationship between prosodic features and their pragmatic functions, attempting to work out how to combine them together into teaching in order to teach students to communicate effectively.

Enhancement of Voice Quality and Bit-rate Reduction by Prosodic Control in G.723.1 Vocoder

Speech coding techniques have been studied not truly to reduce the complexity and bit rate but also to improve the sound quality. CELP type vocoder, used as standard, supports the great stead quality even low bit rate. In this paper, the preprocessing of input speech to reduce the bit rate is different from the conventional vocoder. Different kinds of parameter are used for the preprocessing compared with the other parameters to t＇md the more appropriate parameter for the vocoder. The Parameters are used to synthesize the speech not to encode or decode for coding technique so we proposed the simple algorithm not to have the influence on the processing time or the computation time. The parameters in the preprocessing step are speaking rate, duration, and PSOLA technique.

Identification of Question and Non-Question Segments in Arabic Monologues Using Prosodic Features: Novel Type-2 Fuzzy Logic and Sensitivity-Based Linear Learning Approaches

In this paper, we extend our previous study of addressing the important problem of automatically identifying question and non-question segments in Arabic monologues using prosodic features. We propose here two novel classification approaches to this problem: one based on the use of the powerful type-2 fuzzy logic systems (type-2 FLS) and the other on the use of the discriminative sensitivity-based linear learning method (SBLLM). The use of prosodic features has been used in a plethora of practical applications, including speech-related applications, such as speaker and word recognition, emotion and accent identification, topic and sentence segmentation, and text-to-speech applications. In this paper, we continue to specifically focus on the Arabic language, as other languages have received a lot of attention in this regard. Moreover, we aim to improve the performance of our previously-used techniques, of which the support vector machine (SVM) method was the best performing, by applying the two above-mentioned powerful classification approaches. The recorded continuous speech is first segmented into sentences using both energy and time duration parameters. The prosodic features are then extracted from each sentence and fed into each of the two proposed classifiers so as to classify each sentence as a Question or a Non-Question sentence. Our extensive simulation work, based on a moderately-sized database, showed the two proposed classifiers outperform SVM in all of the experiments carried out, with the type-2 FLS classifier consistently exhibiting the best performance, because of its ability to handle all forms of uncertainties.

Melody- Usul - Poetic Prosodic Meter Relations in Ottoman-Turkish Music

Language, literature, customs and traditions, music and art are cultural items that were transmitted from generation to generation throughout history. In this context, literature is an important source of music culture that takes inspiration from the customs and traditions of a society. Prosodic meter is echoed in form, usul and general structure in works composed from the divan literature and almost lives in the work. In the same way, when examples of folk literature composed by composers and performed by poets and a^lks are examined, it is observed that there are parallels between literary features and form, structure and rhythmic features. The aim of this paper is to reveal the integral link between Melody-Usul and Meter in Ottoman Turkish Music

A Contrastive Analysis of the Structure of Prosodic Words in Mandarin and English

This paper, particularly focusing on the pitch of prosodic words,has conducted a contrastive study on the structure of prosodic words in Englishand Mandarin . This paper reports a Mandarin monologue speech corpus-study, anexperimental phonetic attempt to conduct a study on the pitch of trisyllabic prosodicwords in Mandarin monologue. In addition, taking the characteristics of Englishprosodic words into consideration, the paper makes a contrastive analysis of prosodicwords in English and Mandarin. This study finds that the pitch of trisyllabic prosodicwords in Mandarin is inevitably affected by structural factors. As far as the leftsyllable is concerned, the grammatical category, prosodic hierarchical boundary andthe position of the intonational phrase where the syllable is located, the mid syllableand the right syllable may have influences on the pitch contour of the left syllable.As to the mid syllable, the grammatical category, the left syllable, the right syllableand the position of the intonational phrase where the syllable is located may haveinfluences on the pitch contour of the mid syllable. As for the right syllable, theprosodic hierarchical boundary where the syllable is located and the mid syllable mayhave effects on the pitch contour of the right syllable. Different from the previousfindings of the study on read corpus, this study shows that the mid syllable not onlyhas dissimilatory effects but also has assimilatory effects on the pitch of its precedingsyllable. The left syllable has anticipatory effects on the onset pitch of the mid syllableand the right syllable has coarticulation effects on the offset pitch of the mid syllable.

Prosodically Rich Speech Synthesis Interface Using Limited Data of Celebrity Voice

To enhance the communication between human and robots at home in the future, speech synthesis interfaces are indispensable that can generate expressive speech. In addition, synthesizing celebrity voice is commercially important. For these issues, this paper proposes techniques for synthesizing natural-sounding speech that has a rich prosodic personality using a limited amount of data in a text-to-speech (TTS) system. As a target speaker, we chose a well-known prime minister of Japan, Shinzo Abe, who has a good prosodic personality in his speeches. To synthesize natural-sounding and prosodically rich speech, accurate phrasing, robust duration prediction, and rich intonation modeling are important. For these purpose, we propose pause position prediction based on conditional random fields (CRFs), phone-duration prediction using random forests, and mora-based emphasis context labeling. We examine the effectiveness of the above techniques through objective and subjective evaluations.

基于韵律的英语介词短语挂靠歧义消解研究

在言语交际过程中,说话人口语表达的韵律线索对交际双方的语言理解发挥着重要作用。为了深入了解中国英语学习者英语口语产出的韵律线索与预期表达含义之间的相关性,本文以中国大学英语学习者为研究对象,采用基于情境的半开放式口语产出任务,考察在特定的情境化实验条件下,被试对英语介词短语挂靠造成的歧义句的朗读任务完成情况。研究结果表明:中国大学英语学习者具有对口语韵律特征与句法结构之间关系的潜在敏感性,能够通过口语的韵律停顿和音高重音实现对英语介词短语动、名词挂靠引起的歧义的消解。本研究结果进一步证实了英语口语韵律特征与预期含义之间的相互关系,凸显了韵律训练在大学英语课堂教学中的重要性,为高校英语听说及阅读教学实践的开展提供了参考建议。

自然话语中致歉语“不好意思”的韵律—语用界面研究

致歉语是一种常见的礼貌语用表达,现有研究主要探讨其词句结构和文化语境,较少关注其韵律特征。本文以汉语自然话语中口语致歉语“不好意思”为例,采取语料库研究法,探究该致歉语的韵律特征与语用功能的关系。研究发现,具有不同语用功能的“不好意思”在停顿、时长、音高和音强等方面有着不同的韵律表现,其语用功能和韵律特征之间存在着一定的对应关系。具体表现为:实现概念功能的“不好意思1”与其前后话语成分之间通常无停顿,其平均时长最短,平均音高值和平均音强值都最小。实现礼貌性言语行为功能的“不好意思2”与其前后话语成分之间有时有停顿,有时无停顿,其平均时长居中,平均音高值和平均音强值都最大。实现元话语功能的“不好意思3”与其前后话语成分之间通常存在停顿,其平均时长最长,平均音高值和平均音强值都居中。

融合韵律特征的诗歌生成模型

诗歌生成中的韵律规范和主题一致性一直以来都是自然语言生成领域的研究热点。为提升诗歌生成中的韵律规范,提出了基于Transformer结合韵律特征的诗歌生成模型(Transformer and prosodic features poetry generation model,TPPG)。根据韵律特征建立平仄韵律词库和平声韵脚词库,在Transformer编码器中引入平仄韵律编码,模型训练过程中可以捕获更多平仄韵律特征的信息,学习到多种诗歌韵律;最终根据建立的平声韵脚词库规范诗歌生成韵脚,运用极大后验概率对于候选的诗歌选择当前赋有韵律特征规范的最优诗句,整体提升诗歌规范性和流畅性。实验结果表明TPPG模型生成的诗歌能够很好地符合韵律,在人工评价和机器评价中均有提高。

羌语的韵律构词

不同语系的语言中普遍存在韵律构词机制,词不仅是词法的产物也受到韵律的制约。但羌语的韵律构词规则尚未受到关注。立足前人对羌语各方言丰富的田野调查,文章从韵律语法视角入手,探讨羌语韵律词的类型及其对羌语构词的影响,并得到两个重要结论:1)和汉语、景颇语类似,羌语的韵律词由双音节音步实现,这反映了汉藏语系语言音节计数的特点;2)羌语的韵律词影响和控制了羌语的构词,韵律词是羌语复合构词、紧缩构词和凑补构词的前提和基础。这对羌语的构词、羌语和羌语支的语系定位以及汉藏语系的类型学研究都有重要的意义和价值。

结合轻量卷积的非自回归语音合成方法

对如何有效捕捉音素之间的关联及如何合成韵律丰富的音频进行研究,提出一种结合轻量卷积的非自回归语音合成模型LCTTS。引入轻量卷积建立起音素之间的联系,解决发音出错问题。通过添加音高和能量预测器预测生成语音的韵律,解决音频韵律缺乏问题。训练模型获取梅尔频谱,结合预先训练好的声码器转化为音频。实验结果表明,提出的LCTTS模型优于先前提出的SpeedySpeech模型,在Emotional Speech Database数据集上平均意见得分获得2.8%的提升,梅尔倒谱失真测度下降0.15。

情感语音合成中的语义及韵律特征嵌入方法

针对当前的情感语音合成方法存在合成音频容易忽略文本语义信息的问题,在文本编码器中引入BERT预训练模型,辅助编码器捕获文本语义特征,并提出了语义及韵律特征嵌入方法。缅甸语情感语料的缺乏导致模型难以合成高质量情感语音,因此,文中通过微调各个网络模块参数的方法探索缅甸语情感语音合成模型的训练方法。实验结果表明,文中提出的特征嵌入方法以及训练方法在情感语料缺乏情况下仍能合成出高质量的情感语音,平均情感意见得分分别为4.16与4.18。

新闻播音中韵律边界的声学特性及交际实现

韵律边界是口语交际互动的重要线索,其感知高度依赖于声学线索。本文以《新闻联播》的播读语料为研究对象,采用声学手段对韵律边界音节进行系统研究。结果显示:韵律边界音节的时长均大于非边界音节,而音高和音强则小于非边界音节。此外,韵律边界位置上的阳平和上声存在挤喉音。调音方面,男性和女性的共鸣效果好,高频能量均有所增强,且男性存在演讲者共振峰。本文探究了新闻播音的韵律特性,研究成果可为指导播音教学实践提供一定参考。

汉语拼音输入中韵律边界等级对相邻字母输入时间间隔的影响

由于同音字多,拼音输入不区分音调,导致拼音字母到汉字的转化并不唯一,从而降低了拼音输入法的输入效率。本研究基于汉语输入中的心理运动过程,探讨韵律边界对拼音字母输入时间间隔的影响。为此,我们分别采用自然篇章和歧义拼音字符串进行了两个实验,让被试以全拼方式输入指定汉语文本对应的拼音,记录相邻拼音字母的输入时间间隔。结果表明,拼音的韵律边界等级影响了拼音字母的输入时间间隔,等级越大,则输入时间间隔越长。

烟台话连读变调的韵律辖域

烟台话两字组连读变调是在北方官话字调基础上的语境变调,三字组连读变调却是在吴语词调基础上的模板变调。基于韵律-句法交互作用和韵律音系学理论,对烟台话连读变调韵律辖域的分析显示,韵律-句法匹配原则和韵律标记性制约条件共同作用下形成的最小韵律短语(ωmin),即为烟台话发生连读变调的韵律辖域。依据匹配理论,烟台话连读变调的韵律辖域呈现递归性特征。烟台话连读变调韵律辖域的划分遵循韵律标记性制约条件序列高于匹配原则限制条件的排列规则。制约原则的引入简化了韵律辖域的界定模式,对分析以烟台话为代表的胶东方言具有普遍意义。

“没”字句否定辖域和否定焦点的韵律实现

以“没”后是状中结构的双项动词性成分的否定句为例,考察否定辖域和否定焦点的韵律实现问题。语音实验结果表明:“没”的否定辖域是其后的所有成分,在语调中最突出的投射是辖域内的调域压缩。自然焦点否定句中,说话人优选否定词后毗邻成分为否定焦点,和“没”共同构成自然焦点否定句的联合焦点。否定焦点和否定句焦点可以分离,可以重合或部分重合,由否定辖域内结构和强调焦点的有无及位置决定。

二语语音研究的韵律转向

综合运用文献计量可视化分析和传统文献综述的方法,对国内和国际二语语音的研究动态进行了分析,同时对国际二语语音教学研讨会(PSLLT)近20年的研究议题进行了研究。研究发现:早期二语语音研究主要关注元音、辅音等音段层面,2010年前后呈现出由音段层面转向韵律层面的显著趋势,主要涉及可理解性、韵律、语调、流利度、时长、重音等。通过对二语习得国际权威期刊《二语习得研究》专刊《二语习得中的韵律》栏目中7篇研究论文进行综述,探讨了二语韵律研究的具体做法。

不完全匹配的语音和文本语句级对齐

语音文本自动对齐技术广泛应用于语音识别与合成、内容制作等领域,其主要目的是将语音和相应的参考文本在语句、单词、音素等级别的单元进行对齐,并获得语音与参考文本之间的时间对位信息.最新的先进对齐方法大多基于语音识别,一方面,准确率受限于语音识别效果,识别字错误率高时文语对齐精度明显下降,识别字错误率对对齐精度影响较大;另一方面,这种对齐方法不能有效处理不完全匹配的长篇幅语音和文本的对齐.该文提出一种基于锚点和韵律信息的文语对齐方法,通过基于边界锚点加权的片段标注将语料划分为对齐段和未对齐段,针对未对齐段使用双门限端点检测方法提取韵律信息,并检测语句边界,降低了基于语音识别的对齐方法对语音识别效果的依赖程度.实验结果表明,与目前先进的基于语音识别的文语对齐方法比较,即使在识别字错误率为0.52时,该文所提方法的对齐准确率仍能提升45%以上;在音频文本不匹配程度为0.5时,该文所提方法能提高3%.