On SNR Estimation by the Likelihood of near Pitch for Speech Detection

Authors: Young-Hwan Song, Doo-Heon Kyun, Jong-Kuk Kim, Myung-Jin Bae

Abstract:

People have the habitual pitch level which is used when people say something generally. However this pitch should be changed irregularly in the presence of noise. So it is useful to estimate SNR of speech signal by pitch. In this paper, we obtain the energy of input speech signal and then we detect a stationary region on voiced speech. And we get the pitch period by NAMDF for the stationary region that is not varied pitch rapidly. After getting pitch, each frame is divided by pitch period and the likelihood of closed pitch is estimated. In this paper, we proposed new parameter, NLF, to estimate the SNR of received speech signal. The NLF is derived from the correlation of near pitch periods. The NLF is obtained for each stationary region in voiced speech. Finally we confirmed good performance of the estimation of the SNR of received input speech in the presence of noise.

Keywords: Likelihood, pitch, SNR, speech.

Digital Object Identifier (DOI): doi.org/10.5281/zenodo.1084620

References:

[1] J.S. Han, Speech Signal Processing. Seoul: Osung Media, 2000. ch . 2.
[2] A. Caruntu, G. Toderean, A. Nica, "Automatic Silence/Unvoiced/Voiced Classification of Speech Using a Modified Teager Energy Feature," WSEAS, pp. 62-65, Nov., 2005
[3] WangRae Jo, JongKuk Kim, and Myung Jin Bae, "A Study on Pitch Detection in Time-Frequency Hybrid Domain," Springer-Verlag, Lecture Notes in Computer Science, vol. LNCS 3406, pp.437-440, Feb 2005.
[4] Hans Werner Strube , "Determination of the instant of glottal closure from the speech wave," J., Acoust., Soc., Am, Vol. 5, No. 5, pp. 1625-1629, November 1974.
[5] J.K. Kim, D.S. Na, M.J. Bae "On a pitch alteration technique in transformation domain of speech signals," DCDIS, 2007. pp. 522-526
[6] M. Bae, J. Rheem, and S. Ann "A Study on Energy Using G-peak from the Speech Production Model," KIEE, Korea, Vol. 24, No. 3, pp. 381-386, May 1987.
[7] L. R. Rabiner and R. W. Schafer, Digital Processing of Speech signals. New Jersey: Englewood Cliffs, Prentice-Hall, 1978, ch 6, 7.
[8] P. E. Paparnichalis, Practical Speech Processing. New Jersey: Prentice-Hall, Inc, Englewood Cliffs, 1987.
[9] S. Seneff, "Real Time Harmonic Pitch Detection," IEEE Trans. Acoust. Speech, and Signal Processing, Vol. ASSP-26, pp. 358-365, Aug. 1978.
[10] S. D. Stearns & R.A. David, Signal Processing Algorithms. New-Jersey: Prentice-Hall, Inc, Englewood Cliffs, 1988.
[11] M.J. Bae, "On Detecting the Steady State Segments of Speech Waveform by using the Normalized AMDF," IEEK, Vol.14, No.1, pp. 600-603, June, 1991.
[12] S. J. Kim. "A Segmentation Algorithm of the Connected Word Speech by Statistical Method," IEEK, Vol.26, No. 4, pp. 151-162, Apr., 1989.
[13] M. J. Bae, and S. Ann, "Fundamental Frequency Estimation of Noise Corrupted Speech Signals Using the Spectrum Comparison," J., Acoust., Soc., Korea, Vol. 8, No. 3, June 1989.