20.500.12556/DKUM-98007 Razvoj lokalnega glasovnega asistenta za upravljanje računalnika v slovenskem jeziku Development of a local voice assistant for computer management in slovenian V diplomskem delu smo razvili prototip lokalnega glasovnega asistenta za okolje Windows, ki omogoča upravljanje računalnika v slovenščini in deluje brez internetne povezave. S pomočjo modelov Wav2Vec 2.0 in CNN14 smo raziskali uspešnost klasifikacije omejenega nabora glasovnih ukazov, naučenih na posnetkih enega samega govorca. Pokazali smo, da za zanesljivo prepoznavo ukazov in budne besede ne potrebujemo ogromnih količin podatkov, saj smo prepričljive rezultate dosegli že z manjšim naborom lastnih posnetkov. In our thesis, we developed a prototype of a local voice assistant for the Windows environment, which enables computer control in Slovenian and works without an internet connection. Using the Wav2Vec 2.0 and CNN14 models, we investigated the effectiveness of classifying a limited set of voice commands learned from recordings of a single speaker. We showed that we do not need huge amounts of data for reliable command and wake word recognition, as we achieved convincing results with a smaller set of our own recordings. Prepoznava govora glasovni asistent globoko učenje Speech recognition voice assistant deep learning true true false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2026-05-07 15:42:11 2026-05-29 14:43:28 2026-05-30 03:17:56 0000-00-00 00:00:00 2026 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 VS_Zaloznik_Tine_2026.pdf VS_Zaloznik_Tine_2026.pdf 1 3FCDF0094D7785AD1C7A099E105E1B0F 5dfafe1d85f39a85f1329590b6adff297decf8aedab84473cce6bebb231e904a 165db25d-4de6-11f1-b269-00155d0001a1 https://dk.um.si/Dokument.php?lang=slv&id=206033 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0