prikaz prve stranice dokumenta Prepoznavanje glasa algoritmima za obradu signala
  Download
PDF 1.87 MB
master's thesis
Prepoznavanje glasa algoritmima za obradu signala
2019. urn:nbn:hr:164:076078
Stipinović, Karlo
University of Split
Faculty of Maritime Studies
Department of maritime electr. and. inform. technologies

Cite this document

Stipinović, K. (2019). Prepoznavanje glasa algoritmima za obradu signala (Master's thesis). Retrieved from https://urn.nsk.hr/urn:nbn:hr:164:076078

Stipinović, Karlo. "Prepoznavanje glasa algoritmima za obradu signala." Master's thesis, University of Split, Faculty of Maritime Studies, 2019. https://urn.nsk.hr/urn:nbn:hr:164:076078

Stipinović, Karlo. "Prepoznavanje glasa algoritmima za obradu signala." Master's thesis, University of Split, Faculty of Maritime Studies, 2019. https://urn.nsk.hr/urn:nbn:hr:164:076078

Stipinović, K. (2019). 'Prepoznavanje glasa algoritmima za obradu signala', Master's thesis, University of Split, Faculty of Maritime Studies, accessed 05 August 2022, https://urn.nsk.hr/urn:nbn:hr:164:076078

Stipinović K. Prepoznavanje glasa algoritmima za obradu signala [Master's thesis]. Split: University of Split, Faculty of Maritime Studies; 2019 [cited 2022 August 05] Available at: https://urn.nsk.hr/urn:nbn:hr:164:076078

K. Stipinović, "Prepoznavanje glasa algoritmima za obradu signala", Master's thesis, University of Split, Faculty of Maritime Studies, Split, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:164:076078

Please login to the repository to save this object to your list.
Metadata
TitlePrepoznavanje glasa algoritmima za obradu signala
Title (english)Speech recognition by using signal processing algorithm
AuthorKarlo Stipinović
MentorJoško Šoda (mentor)
Committee memberIgor Vujović (predsjednik povjerenstva)
Committee memberDanko Kezić (član povjerenstva)
Committee memberIvan Pavić (član povjerenstva)
GranterUniversity of Split
Faculty of Maritime Studies
(Department of maritime electr. and. inform. technologies)
Split
Defense date and country2019-09-30, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Traffic and Transport Technology
Abstract
U radu se opisuje izrada programa za klasifikaciju izgovorenih riječi „Marvin“, „Yes“ i „No“ pomoću neuronske mreže sa dugoročnim pamćenjem (engl. Long Short Term Memory - LSTM) u programskom jeziku Python. Kao uvod u obradu i analizu audio signala opisuju se pojmovi uzorkovanja, Fourierovog reda i transformacije te diskretne Fourierove transformacije. Dobivanje korisnih informacija uz istovremeno smanjenje količine podataka ostvareno je pomoću sljedećih svojstava zvučnih signala: stopa prelaska nule, kepstralni koeficijenti po Mel-skali, spektralni centroid i spektralno opadanje. Za detaljni pregled preciznosti klasifikacijskog modela proučavalo se njegove funkcije greške i preciznosti, te klasifikacijsko izvješće i konfuzijsku matricu.
Abstract (english)
This thesis describes development of a program for classification between spoken words „Marvin“, „Yes“ and „No“ using Long Short Term Memory (LSTM) neural network in Python programming language. As an introduction to the area of audio signal analysis and processing, description of following terms is provided: Fourier series, Fourier transform and discrete Fourier transform. Features that are used as a measure of acquiring useful information and data reduction of audio signal are the following ones: zero crossing rate, Mel-Frequency Cepstral Coefficients, spectral centroid and spectral rolloff. Precision and loss functions with classification report and confusion matrix are shown as a measure of detailed overview of classification model accuracy.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:164:076078
Study programmeTitle: Maritime Electrotechnical Technologies
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehničkih tehnologija (magistar/magistra inženjer/inženjerka elektrotehničkih tehnologija)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2019-12-11 09:58:19