Statistical analysis of speech signals by weighted ASMDF
Roudra Chakraborty, Sagnik Sinha, Debapriya Sengupta · International Conference on Signal Processing · 2010
In the previous two papers,[1] and [2], tracking of fundamental frequency of voiced signals in clean and noisy environment by ASMDF have been demonstrated and compared with two classical methods autocorrelation and AMDF. It has been shownn that ASMDF gives near error-free results in case of clean signal. But in case of noisy signals, the percentage of relative errors have increased. Also weighted autocorrelation method [22] gives somewhat better results than ASMDF. In this paper, we propose a new method which is derived by refining ASMDF and show that the results using this method are better than ASMDF and weighted autocorrelation.