Prepoznavanje hrvatskih riječi pomoću Nao robota

master's thesis

• Title (english): Recognition of Croatian words with NAO Robot
• Author: Matija Bagić
• Mentor: Stjepan Bogdan (mentor)
• Committee member: Stjepan Bogdan (predsjednik povjerenstva)
• Committee member: Zdenko Kovačić (član povjerenstva)
• Committee member: Ivana Palunko (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2014-07-16, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu osmišljen je sustav za prepoznavanje glasovnih naredbi za NAO robota. Implementirane su glavne komponente takvog sustava. Osmišljen je i implementiran sustav za prikupljanje glasovnih naredbi u formi web aplikacije. Uspoređena su dva postupka ekstrakcije značajki iz zvučnih zapisa, te su mel-frekvencijski kepstralni koeficijenti odabrani zbog manje vremenske složenosti. Odabrana su tri klasifikatora - stroj s potpornim vektorima, k najbližih susjeda te nasumična šuma. Sva tri klasifikatora evaluirana su na istom skupu primjera, te je zaključeno kako najbolje performanse pokazuje stroj s potpornim vektorima. Cijeli sustav implementiran je u programskom jeziku Python uz korištenje više pomoćnih modula.

Abstract (english)

The paper describes a speech-recognition system for the NAO robot. The main components of such a system have been developed and implemented. The system is comprised of a speech recording web application, feature-extraction and classification. Linear predictive coding (LPC) and mel-frequency cepstral coefficients (MFCC) have been compared as two possible feature-extraction procedures, and the MFCC was chosen as the faster one. Three classification algorithms have been compared - k Nearest Neighbours (kNN), Support Vector Machine (SVM) and Random Forest, and SVM was chosen for the highest prediction accuracy. The whole system was implemented in the Python programming language, using a number of modules.

• Keywords: prepoznavanje govora; Nao robot; k najbližih susjeda; stroj s potpornim vektorima; nasumična šuma; linearno prediktivno kodiranje; mel-frekvencijski kepstralni koeficijenti
• Keywords (english): speech recognition; Nao robot; k nearest neighbors; support vector machine; random forest; linear predictive coding; mel-frequency cepstral coefficients
• Language: croatian
• URN:NBN: urn:nbn:hr:168:977401
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 07:41:06