Study on key-to-pinyin conversion problem for Chinese phonetic input technique on digit keyboard;

In order to detect the length of secret messages embedded by audio-to-image audio steganography,via researching distribution characteristic of wavelet coefficients histogram after audio-to-image transform the peak frequency feature which has linear relationship with the embedding capacity is extracted,and the linear regressive model is established.

A survey of voice conversion and its relevant technology;

Research of the Chinese Voice Conversion System;

Research and Implementation of Voice Conversion Techniques Based on ARM9;

Experiment is carried out on voice conversion of 5 vowels from international phonetic symbol and speech.

voice conversion technique aims to modify the source speaker s speech to make it sound like a designated target speaker’s speech, of which the spectral envelope mapping algorithm is the key part.

The application of three different regression methods which named Polynomial Regression, Linear Multi-variant Regression and Support Vector Regression in voice conversion is compared in this paper.

In order to overcome the difficulty in fusing more features into n-gram,a Pinyin-to-Character conversion model based on Support Vector Machines(SVM) is proposed in this paper,providing the ability of integrating more statistical information.

It also shows that the CFS based language model has lower model perplexity and higher pinyin-to-character conversion correctness compared with the model based on previous CFS extraction method.