| Title |
Fractal modeling of speech signals / |
| Translation of Title |
Fraktalinis šnekos signalo modeliavimas. |
| Authors |
Kizlaitienė, Ieva |
| Full Text |
|
| Pages |
46 |
| Keywords [eng] |
Fractal dimension, Speech emotion ; Non-linear ; Katz's method, Castiglioni's method, Higuchi's method ; Neural network ; Feature evaluation. |
| Abstract [eng] |
Fractal dimension is an important characteristic of fractals that describes the degree of fractals' irregularity over multiple scales. In this paper we develop five new methods for fractal dimension calculation and analyze their theoretical and practical applications. Theoretical research is performed on extreme cases, harmonic function and Weierstrass sine function. Practical experiment uses fractal dimension-based features for speech emotion classification task. All results are compared with Katz, Castiglioni and Higuchi algorithms. The study suggests that in most cases proposed methods' performance is better and more reliable and that the new methods can be used in future researches. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2021 |
