耳语音

Reconstruction of normal speech from Chinese whispers can be applied in several fields, such as the private speech communication in public, reconstruction of normal speech for the aphonic individuals, the special need for the forensic work and etc.

耳语音转换为正常音，可应用于公共场所的通信、失语者的语音恢复和司法工作的某些特殊需要等方面。

In this method, we use the pooled correlogram based on human auditory model to extract the pitch of speech. An unsupervised lateral inhibitory network is used to get the peak position, which simulates the lateral inhibitory phenomenon in human auditory system.

采用基于人耳听觉模型的相关图来提取语音信号的基频，运用无监督的侧抑制神经网络来模拟人耳侧抑制属性进行基频检测，为了克服在低信噪比情况下侧抑制神经网络的误判问题，引入了相邻语音帧的语音基频的帧间约束。

And in some professional fields whisper is also useful, the linguistician may study people's perceptual mechanism through whisper, the normal speech can be reconstructed for the patient whose larynx is cut, and the speaker can be identified through whispers.

耳语在一些专业领域同样有着重要的应用意义，例如语音学家用耳语音研究语音感知问题、无喉病人的语音恢复问题、公安和法院可通过耳语音分析来辨别说话人身份等等。

In this thesis, we design such a monaural speech segregation system based on an auditory perceptual model.

本论文中，我们设计一个基於听觉感知模型的单耳语音分离系统。

Various spectral-temporal cues extracted from the model are used for monaural speech segregation. Then, a self-organizing feature map neural network is utilized to mimic the neural function in segregating and clustering a mixed sound into separated sounds.

我们从此模型中取出不同在时域-频域上的一些使用於单耳语音分离系统的线索，之后，利用自组织映射图来模拟神经元将混合的语音分组和归类成分开的语音。

Some information such as pitch and harmonic of each speech stream, interaural time difference and interaural intensity difference and etc could be obtained from overlapping speech by the analyzing and modeling of overlapping speech and noisy speech, based on which the interested speech can be reconstructed and overlapping speech can be separated.

通过对混叠语音和含噪语音进行分析建模，对混叠语音分别提取其各自语音流的基音及谐波、双耳时间差ITD和强度差IID等信息，并以此重构目标语音分量，达到分离混叠语音的目的。

The most important part of an electronic cochlea system is the multiple-channel receiver-stimulator implemented in this thesis, which will be implanted into the mastoidal cavity nearby the cochlea. Data signals transmitted by an external speech processor are received through a tuned inductive coil. The power supply for the receiver-stimulator is derived from the rectified and filtered data signals. Meanwhile, the clock and the data envelope are extracted.

本文实现的多道（22根电极）电子耳蜗接收刺激器是人工电子耳蜗系统中的重要组成部分，它将直接被置入患者耳蜗附近的乳突骨腔中，通过电感耦合接收从体外语音处理器发射的数据信号，通过整流滤波为接收刺激器提供电源，同时提取时钟信号和数据包络[31]。

The formants extraction of whisper speech is more difficult than that of normal speech because of the difference of the two kinds of speech.

耳语音不同于正常语音的特性，使得准确提取其共振峰存在更大的困难。

The authors analyze the questions of pole interaction while extracting formants. A new LPC improved algorithm based on pole interaction by arranging their order of poles is proposed in this paper. The experimental results show that the new algorithm can solve the two questions and it can extract the first three or four formants of whisper speech.

本文分析了共振峰提取中存在的极点交叉问题，通过对极点排序，提出基于极点交叉的LPC改进算法，从而有效解决这两个问题，实验结果证明此算法能够根据实际要求准确提取耳语音的前三个或前四个共振峰。

So whisper has no vocal fold vibration, no pitch and its energy is much lower than that of normal speech.

由此可以确定耳语音的主要特点是发音时音量低且声带完全不振动，即没有基频。

As she looked at Warrington's manly face, and dark, melancholy eyes, she had settled in her mind that he must have been the victim of an unhappy attachment.

每逢看到沃林顿那刚毅的脸，那乌黑、忧郁的眼睛，她便会相信，他一定作过不幸的爱情的受害者。

Maybe they'll disappear into a pothole.

也许他们将在壶穴里消失