Visiting fellowship in Signal Processing of Nonstationary Signals

多年来，许多研究人员在信号处理应用，如语音处理，一直在努力寻找紧凑和易于处理的波形表示。鉴于许多信号的非平稳性质，已经提出了许多模型试图捕捉时变特征。虽然估计时变计算模型的想法并不是一个新的想法，但最近发表的研究报告表明，人们明显倾向于关注高时间分辨率估计的参数（瞬时参数）。本研究结果可应用于文本到语音系统的开发，产生高质量的语音。低比特率语音编码。基于知识的语音识别。多媒体系统的语音压缩。

For years many researchers in the signal processing applications,such as speech processing, have strived to find compact andeasy-to-handle representations of the waveform. Given thenon-stationary nature of many signals many models have been proposedattempting to capture time-varying features. Although the idea ofestimating time-varying computational models is not a new one, recently published research reports indicate a marked tendency to focus on parameters estimated with high temporal resolution (instantaneous parameters). The results obtained with this research could be applied in the development of :Text to speech systems producing high quality speech.Low bit rate speech coding.Knowledge-based speech recognition.Speech compression for multimedia systems.