语音

Instead of statistical description, a feature subspace with the base of some significant eigen vector extracted from covariance matrix is constructed to describe speech feature distribution of speaker. Distance metrics for measuring distance between input feature vector and subspace are also proposed for pattern matching.

说话人语音训练样本提取特征后在语音特征观察空间形成具有一定散度的分布，根据主元分析原理和分布散度提取主要散度本征向量作为基底构成说话人语音特征子空间，并通过测试语音特征矢量与子空间的距离测度进行模式匹配。

Results: Several initial iterations of the paradigm failed to produce useful activation maps and required modification of the protocol to reduce the complexity of language tasks, eliminate verbalization and implement head immobilization to reduce corruption of data from involuntary head movements.

结果：所有受试者均按照要求完成全部实验，语音相近组平均反应时间为1.214±0.167ms，语音不相近平均反应时间为1.215±0.135ms；语音相近组平均判断正确率为84.3％±13.5％，语音不相近组平均判断正确率为94.1％±8.0％。

For practical systems it is necessary to make speech recognition systems robust to interfering noise, different speaker and easy confusable phonemes.

系统的鲁棒性问题是语音识别实用化的关键，它包括非特定人语音识别，混淆词语音识别和噪声环境下的语音识别三个基本问题。

This article, on the basis of English phonetic features, discusses the effectiveness of modern technology in the teaching and learning English phonetics, particularly the application of multi-media and speech spectrographic devices to the phonetic teaching.

基于英语语音特征及其教学方法，本文探讨现代技术在语音教与学中的效用，特别是多媒体和语音摄谱仪在语音教学中的应用。

In speech communication, almost half of the time is the silence state, which contains much redundant information need to be taken away to improve code efficiency, so the silence frame should be compressed; On the other hand, to satisfy people's listening feelings, speech code implemented on UWB would be a new future's research direction on speech code. Nowadays the compression on narrow band or wideband have been studied a lot, but the silence compression on UWB need to be studied more, which is the main job of this paper.

在语音通信中，平均一半的时间处于无语音的空闲状态，这里蕴含着大量的冗余信息，所以去掉这些冗余信息将会大幅度的提高语音编码的传输效率，这就需要对静音信号进行压缩；此外，为了满足人们听觉享受，在高带上实现语音编码也将成为未来的研究方向，现如今窄带和宽带的静音压缩技术已经较为成熟，但高带的静音压缩技术尚待发展，这也正是本文研究的必要性。

The speech semaphore analogs wave form and delivers a speech through IP method and ignore to is a real time appliance a traffic still a non- real time applied traffic, the pass looks shore wants to steer a simulation data convert to the speech semaphore first, is also vs simulation the speech semaphore steer 8 or 6 quantizes, then send go into to buffer store ward amid, the dimension of dampener can be sur the request select of the delay and the encode.

语音信号是模拟波形，通过IP方式来传输语音，不管是实时应用业务还是非实时应用业务，道貌岸首先要对语音信号进行模拟数据转换，也就是对模拟语音信号进行8位或6位的量化，然后送入到缓冲存储区中，缓冲器的大小可以根据延迟和编码的要求选择。

In 1999 the U.S. began to introduce and promote the ISD voice circuits, are widely used for advertising, product selling machine, recording key chain, telephone response tape recorders, voice alarm systems, toys and other voice recording electronic products, and in 2001 and Hong Kong Ka Ling Electronics, Winbond Electronics to promote a series of voice products and four SCM.

1999年开始引进并推广美国ISD语音电路，产品广泛应用于广告宣传机、产品叫卖机、录音匙扣、电话应答录音机、语音报警系统、录音玩具等多种语音电子产品，并于2001年与香港嘉灵电子合作，大力推广华邦电子的系列语音产品及4位单片机。

Most of the vocoders, such as CELP, MELP are based on this model, which make it better for the applications when less storage space is needed.

语音信号的分析合成采用线性预测模型，线性预测模型是一种在技术上比较成熟的语音模型，大部分的语音编码器如CELP、MELP 等都是基于此模型，所以在需要减小语音库的存贮空间时有较大的优势。

And therefore, their performance is not good. But it is well known that there are intrinsic relations between speech basic units. It is necessary to develop an adaptation algorithm to use the relations and few adaptation data to re-estimate more model parameters and then get speech models more suitable for the new speaker.

最大似然平滑预测方法由于不同语音模型之间存在着相关关系，有必要开发话者自适应技术，利用语音模型之间的关系和少量新话者的语音数据，尽量多地适应模型参数，使得最终的模型更代表新话者的语音特点。

S the computation power and the memory capacity increase, the corpus-base speech synthesis system has become the best and most popular speech synthesis system. Based on the system, the linguistic features are first derived after the text is parsed, then some appropriate units are selected as candidates. Finally, the well-pronounced speech is synthesized by concatenating the best unit sequence by the synthesizer part of system.

bstract：随著计算机之运算速度及记忆体容量的加快与增大，以大型语料库为基础之语音合成系统已成为目前最普遍且品质最好的语音合成系统，其方法是先输入欲合成之文字后，再经过分析得到语言参数，接著在大型语料库中依据所得到之语言参数找到对应之候选合成单元组，最后透过语音合成器挑选出最合适的单元组进行串接以得到合成语音。

Breath, muscle contraction of the buttocks; arch body, as far as possible to hold his head, right leg straight towards the ceiling (peg-leg knee in order to avoid muscle tension).

呼气，收缩臀部肌肉；拱起身体，尽量抬起头来，右腿伸直朝向天花板（膝微屈，以避免肌肉紧张）。

The cost of moving grain food products was unchanged from May, but year over year are up 8%.

粮食产品的运输费用与5月份相比没有变化，但却比去年同期高8％。