[语音合成] FPTTS_demo
语音合成，又称文语转换（Text to Speech）技术，能将任意文字信息实时转化为标准流畅的语音朗读出来，相当于给机器装上了人工嘴巴。
我们的实验室实现的英语语音合成技术音质好，自然度高，可懂度高。语音合成并且整个系统基于C开发，可以很方便的移植到其他移动平台。下一步我们会开发中文语音合成技术，以及在 提高合成语音的自然度、丰富合成语音的表现力、降低语音合成技术的复杂度、多语种文语合成方面做进一步的研究。
我们实验室的博客地址参看 http://blog.sina.com.cn/foxpigeontech
Speech synthesis, also known as TTS (Text to Speech) technology, can any text information in real time into a standard smooth voice read out, which is equivalent to a machine fitted with an artificial mouth.
Our laboratory English speech synthesis technology to achieve good sound quality, natural high, high intelligibility. Speech synthesis and development of the entire system is based on C, can be easily ported to other mobile platforms. Next, we will develop Chinese speech synthesis technology, as well as in improving the naturalness of synthesized speech, rich and expressive speech synthesis, speech synthesis technology to reduce complexity, multi-lingual text to speech synthesis to do further research. (2013-12-19, Visual C++, 2717KB, 下载9次)
[语音合成] Spedaker_Adapting_in_Speech_recognizing
：自适应技术在近年来得到越来越多的重视，其中应用广泛的包括,-.、,//0，该技术利用少量特定
人数据就可以调整码本，快速地提升识别性能，它要求原始的码本有很好的说话人无关性。本文介绍了结合
,//0 自适应的说话人自适应训练（1234536 -74289:3 649<9<=，以下简称1- ）算法，这种方法将每个说话人码本
视为说话人无关码本经过线性变换的结果，在此基础上训练的说话人无关码本更有效剔除了说话人相关信
息，因此在说话人自适应中时能根据特定数据调整更好地逼近说话人特性，从而有更好的性能表现。
Introduced the term network based maximum likelihood linear regression unsupervised adaptive algorithm, and an improved. According to decode the received word net estimated transformation parameters, the word error rate of net potential is far less than the recognition results, it can make parameter estimation more accurate. A major drawback is that the traditional calculation enormously difficult practical, this paper presents two improved technology: 1 compression using word posterior probability network 2 time information using the word limit state statistic calculation. Experimental determination of the relative error rate than traditional down. (2010-08-13, PDF, 215KB, 下载5次)
[语音合成] New_method_for_speech_endpoint_detection_under_low
：Duffing振子相变对与参考信号频差较小的周期小信号具有敏感性，对白噪声和与参考信号频差较大的周期干扰信号
具有免疫力。利用语音的准周期性提出了一种基于混沌理论的语音端点检测新算法，讨论了如何利用Duffing振子的间歇混沌
特性来检测语音端点。将含噪语音作为Duffing振子的外部参考信号的摄动，根据系统相图是否改变来检测语音段和噪声段。
通过尺度变换并设计了振子阵列以检测频率范围较大的语音信号。实验结果表明，基于混沌的语音端点检测算法在低信噪比
下明显优于传统的能量语音端点检测法。
Duffing oscillator phase transform is sensitive to small signal that has tiny ~equency difference with the
referential signal，and immunes against random noise and larger interference signal that has larger frequency differ—
ence with the referential signa1．In this paper，a new method is proposed for endpoint detection based on chaotic the—
ory using the qusi—periodical characteristic of speech．How to test speech endpoint using intermittent chaos of Duffing
oscillator is discussed．There we take noisy speech as the perturbation of the referential signal of Duffing oscillators，
and we can test speech segments and noise segments according to phase trajectory．Oscillator array is designed and
connected with scale transform to detect the speech signal whose frequency range is wide． Experiments show that
chaotic method is superior to traditional energy method obviously under low SNR condition for endpoint detection． (2010-04-24, Others, 258KB, 下载36次)
[语音合成] IVR
交互式语音应答IVR核心源码,交互式语音应答系统提供自动语音服务，是企业为客户提供的自助服务的主要设备。系统采用用户导向的语音目录，根据客户选择（通过电话键盘或语音）完成相应的信息查询和命令执行，所以可以说是通过电话机的按键控制计算机。通过在IVR后端连接数据库，IVR系统能为客户提供动态的实时信息
Interactive Voice Response IVR core source, interactive voice response system provides automated voice services, providing customers with the enterprise of self-service major equipment. System uses a voice-driven directory of users, according to the customer to choose (through the telephone keyboard or voice) to complete the corresponding information and orders of the Executive, so it can be said that through the telephone keypad to control the computer. IVR connected through back-end database, IVR system can provide customers with dynamic real-time information (2007-11-24, Visual C++, 5243KB, 下载191次)