ch1-绪论.pptVIP

第1章 绪论 From prehistory to the new media of the future, speech has been and will be a primary form of communication between humans. Speech will become the next major trend in the personal computer market in the near future. 语音信号产生的数字模型Speech Production Model 语音信号的应用Application 语音编码技术 Speech Coding 语音合成技术 Speech Synthesis 语音识别技术 Speech Recognition 语音增强技术 Speech Enhancement 声源定位技术 Speaker Tracking/ Sound Source Localization Sound card enables the computer to Record sound from a microphone Output sound through a speaker Manipulate sound stored on a disk Sound card contains A/D and D/A converters Sampling rate generator Most sound cards use DMA to read and write audio data Two types of audio file formats: Raw format, no header raw Self-describing format, with a file header Wav Midi Aiff, au, voc Mpeg Rm, ra, ram Audio file parameters Sampling rate Bits per sample Channels THANKS * * * * * * 语音信号处理 Speech Signal Processing 教材：语音信号处理 机械工业出版社 参考书：语音信号数字处理 电子工业出版社 学习的目标： 保障现代通信领域中人与人之间、人与机器之间的信息交换的顺畅和自然。 Representation of speech signals waveform parametric preamplifier anti-aliasing filter A/D conversion Discrete System Speech Excitation 根据人类语音的发音的生理过程和语音信号的声学特征来建立。 ☆短时时域处理技术 ☆短时傅立叶分析 ☆语音信号的线性预测Linear Predication ☆语音信号的同态滤波和倒谱分析 ☆矢量量化技术Vector Quantification ☆隐含马尔可夫模型HMM(Hide Markov Model) 语音信号的处理方法 语音信号是时变信号，只是在一个短时短内可近似看作是平稳信号。 语音信号处理涉及语言学、声学、认知科学、生理学、心理学和数理统计等多学科知识。用数字化的方法进行语音的传送、存储、合成、识别和增强等技术是整个数字化通信网中最重要最基本的组成部分。 Speech Coding Digital Channel The goal is reduce the information rate, measured in bits per second, while maintaining the quality of the original speech waveform. PCM:64kb/s waveform coder LPC-10:2.4kb/s parametric coder GSM:13kb/s hybrid coder 欢迎使用微软中国研究院中文语音合成系统 采用TTS技术合成语音 http://research. M/speech/tts Speech Synthesis 金山词霸/电子词典 从音乐中提取有用信号 Speech Enhancement the goal is to improve the quality of degraded speech. Speech Recognition