PROBABILITY DENSITY FUNCTION OF AVERAGE POWER OF REAL SPEECH SIGNALS
DOI:
https://doi.org/10.35120/kij5403459pKeywords:
Speech signals, Statistical parameters of the signal, Signal processing, Adaptation techniquesAbstract
It is known that many real signals, such as speech signals, are non-stationary processes that express their
feature through changes in parameters over time. By observing speech in shorter time intervals, the property of
stationarity can be notice. This characteristic enables the application of techniques for adaptation to local signal
characteristics in signal processing algorithms. Many of these algorithms are described by standards, and in addition
to intensive development in this area in last decades, there is a constant need for the development of new solutions
and standards. One of the most used parameters of the speech signal for adaptation is the mean (average) signal
power. The change of speech in time results in a wide dynamic range of average power. In addition to the predicted
dynamics of average power, in the design of systems with adaptive techniques it is important to include other
parameters, among which is the Probability Density Function (PDF) of the average power. The goal of the research
presented in this paper is the analysis of the probability distribution of the average power of speech signals, based on
which the adaptability in the development of algorithms in digital processing would be improved, which would
ensure higher quality and less requirements in data transmission and storage. In addition to the theoretical
consideration, the research was conducted on real speech signals of different speakers. In modern technical systems,
where Internet technologies are prominent, processing, transmission and memorization of speech is executed frame
by frame. Therefore, an analysis of the probability density of the average power for different frame lengths was
carried out in the experiment. In the experimental part, for each of the speech signals of the speech corpus, the
Probability Density Function that best describes the average power values per frame was determined. Experimental
research results indicate that the function that best describes the average power is different for different speakers. In
addition, when observing one speaker, the Probability Density Function is different for different frame lengths. It
can be concluded that when it comes to adaptive techniques in the digital processing of the speech signal, it is
important to consider the characteristics of the average power, among which is the Probability Density Function of
the average power
References
Chu, W. C. (2003). Speech Coding Algorithms. New Jersey: Foundation and Evolution of Standardized Coders. John Wiley & Sons, Chapter 5, 143-158.
Gazor S., & Zhang, W. (2003). Speech probability distribution. IEEE Signal Process. Letters, 10(7), 204 207. doi: 10.1109/LSP.2003.813679
Perić, Z., & Nikolić, J. (2012). An adaptive waveform coding algorithm and its application in speech coding. Digital Signal Processing, 22(1), 199-209. doi: 10.1016/j.dsp.2011.09.001
Peric Z., Petkovic G., Denic, B., Stanimirovic, A., Despotovic, V., & Stoimenov, L. (2020). Gaussian Source Coding using a Simple Switched Quantization Algorithm and Variable Length Codewords. Advances in Electrical and Computer Engineering, 20(4), 11-18. doi: 10.4316/AECE.2020.04002
Petković, G. (2015). The Characteristics of the Speech Signal and Application in Si gnal Compression. Sixth International Scientific Conference “Knowledge–The Power of Knowledge”, Agia Triada, Greece, Institute of knowledge management - Skopje, 10(1), 545-548.
Petković G. (2016). Average power and variance in the processing of real voice signals. Tenth International Scientific Conference “Knowledge–The Power of Knowledge”, Agia Triada, Greece, 7-9.10.2016, Institute of knowledge management - Skopje, 14(2), 740-744.
Petković, G., Perić, Z., & Despotović, V. (2017). Switched uniform scalar quantization adapted to mean and variance for speech coding. Facta universitatis - Series: Automatic Control and Robotics, 16(3), 263 274. doi: 10.22190/FUACR1703263P
Petković G. (2021). Autokorelacija odmeraka realnih govornih signala. Niš: Zbornik radova, Akademija tehničko-vaspitačkih strukovnih studija.
Petković, G., Perić, Z., & Stoimenov, L. (2016). Switched Scalar Optimal µ-law Quantization with Adaptation Performed to both the Variance and the Distribution of Speech Signal. Elektronika Ir Elektrotechnika, Kaunas University of Technology, 22(1), 64-67. doi: 10.5755/j01.eee.22.1.14111
Sayood, K. (2005). Introduction to Data Compression. San Francisco: Elsevier Science, Chapter 9, 227-270.
Vucic, N., Peric, Z., & Petkovic, G. (2018). Design of Switched Quantizers and SpeechCoding Based on Quasi-Logarithmic Compandor. Elektronika Ir Elektrotechnika, Kaunas University of Technology, 24(6), 82-86. doi: 10.5755/j01.eie.24.6.22295
