说话者

The basic idea of MLMI is to compute the speaker adapted model of a test speaker by a linear convex combination of a set of speaker dependent models according to maximum likelihood criterion.

其基本思想是，利用语音单元间的相关性，根据最大似然准则由一组说话人相关模型的线性组合得到测试者的说话人自适应模型。

5 I heard a holy one speaking, and another said to whichever one it was that spoke,＂ How long shall the events of this vision last concerning the daily sacrifice, the desolating sin which is placed there, the sanctuary, and the trampled host?

我听见一位圣者在说话，另一位圣者问那说话的说：＂关于废除日常祭，招致毁灭的罪孽，圣所和天军遭蹂躏的异象，要延长到何时呢？

A work of great difficulty to himself, requiring much prudence, zeal and meditation, and to the natural and corrupt man will be very unpleasant ; yet he is to endeavour to perform it in such a manner, that his auditors may fell the word of God to be quick and powerful, and a discerner of the thoughts and intents of the heart; and that, if any unbeliever or ignorant person be present, he may have the secrets of his heart made manifest, and give glory to God.

应用经文必须现实﹕解经者必须留意，有没有使经文向人们实际的情况说话。昨日的应用，不一定能向今天的情况说话。…「若只是指责古老的错误，或过时的生活，做事方式，这种的热诚是卑贱的。我们必须考虑现今一代需要的是什麼。」解经者能适切地应用《圣经》，刺透人心，造就生命（有别於不适切，混乱的应用），无疑是非常不容易的，需要智慧，热心和默想，而对属血气和败坏的人来说是非常不舒服的事（考虑到这一点，可能诱惑上帝的仆人失去控制，因为平常人都喜欢受到爱戴）。可是他必须努力履行这任务，好叫听众能感觉到上帝的话语是大能有力的，使人苏醒的，查验人的心思意念的；若有非信徒或无知的人在场，他们心中隐秘的事就必显露，叫荣耀归给上帝。

This paper presents the methods to improve the performance of mispronunciation detection at syllable level for Mandarin Chinese from two aspects: introducing the speaker adaptive training and the selective maximum likelihood linear regression to get a better acoustic statistical model, and proposing speaker normalization backend because of the limited information and the different rating level for the different pronunciation level.

该文一方面为了获得对于发音错误检测更好的声学模型，引入了说话人自适应训练和选择性最大似然线性回归技术；另一方面，由于字发音检错中存在严重的信息量不足问题和专家对于不同水平说话人的评价标注不一样，在后端上加入了话者得分归一化技术。

In this process, the interpreter should choose the proper context and relate the speaker"s ostensive stimuli with his cognitive environments to infer the speaker"s intention, and then take into consideration the target audiences cognitive environment and reproduce this intention.

在这过程中译者既是信息接受者又是信息传递者，不仅要面对说话人选择相关的语境，把说话人提供的明示信息和自己的认知环境相结合，推理出说话人的意图，又要考虑听话人的认知语境，为他们提供最佳关联性的语言信息来传递说话人的这一意图。

Then I heard a holy one speaking; and another holy one said to the particular one who was speaking, How long will the vision concerning the daily sacrifice and the transgression that desolates apply, so that the sanctuary and the host are trampled down?

8：13 我听见有一位圣者说话，又有一位圣者问那说话的圣者，说，这除掉日常祭祀，和造成荒凉的过犯，以致圣所与军兵遭践踏的异象，要持续多久呢？

Most normally fluent speakers become highly disfluent when exposed to DAF and many stutterers experience decreased stuttering under DAF.

绝大多数说话流畅的正常人当暴露在DAF的环境中时，说话会变得非常结巴；而许多口吃者在DAF的环境中说话时，口吃却会减轻。

With analysis of how to make public persuasion work better, this research is meant to construct a "theory of intimacy," which is based on intersubjectivity by J. Habermas, inter-referring by C. Taylor, imagination by H. Arendt, and sensus communis by G. Vico.

本研究从Habermas的相互主体性、Taylor的相互指涉、Arendt的想像力及Vico的共通感受等概念中，试图建构「亲密关系」的理论架构，探讨说话者如何使公共说服更为可能。

Bush said the signing will begin a new period in relations between the United States and Russia. He said it will make the world more peaceful and put old aggressions in the past.

以上两句主句皆为一般过去时，而宾语从句均为一般将来时态，可能是因为作者认为&该条约的签署将揭开两国关系的新篇章&等具有必然性，或者表示说话者Bush的一种强烈的愿望。

In sociolinguistic studies, speakers are treated as members of social groups.

社会语言学研究中，说话者被当作是社会群体的成员。

No clear conclusions can be made about the predictie alidity of specific measures, they found, except for criteria for the underuse of drugs for cardioascular disease.

除了心血管病不合理用药标准外，对于特定评价方法的预测效度没有明确的结论。

The third and last is that almost all of them, being infinite in number, have been impostures, and by idle and crafty brains merely contrived and feigned after the event past.

第三、即最后一点是：几乎所有这些不胜枚举的预言都是欺人之谈，都是在事件过后由无聊而狡猾的脑筋推测与凭空捏造的。