Izražanje čustev v sintetiziranem govoru

LIPOVŠEK, GAJ (author), Bajec, Marko (mentor)

Področje sinteza zvoka se z napredovanjem umetne inteligence hitro razvija. Ustvarjeje govora je mnogo hitrejše, govor zveni bolj naravno in izgovorjava je lahko prilagojena željam uporabnika. V tej nalogi smo se ukvarjali s sistemom za sintezo govora FastPitch. Pokazali smo kako ga namestiti lokalno in uporabiti. Razvili smo spletno aplikacijo, ki omogoča interaktivno spreminjanje višine tona v ustvarjenem govoru. Učinke spreminjanja višine tona na govoru smo preučili in opisali.

• Language: Slovenian
• Keywords: sinteza govora, teskt v govor, spletna aplikacija, fastpitch, čustva v govoru
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2024
• PID: 20.500.12556/RUL-162601
• COBISS.SI-ID: 214200323
• Publication date in RUL: 25.09.2024

Secondary language

• Language: English
• Title: Expressing emotions in synthesized speech

Abstract

Speech synthesis is quickly developing as the field of artifical inteligence progress. Synthesising speech is much faster, sounds more natural and can be customized for user preference. In this thesis, we explored FastPitch, a speech synthesis system. We showed how to deploy the model locally and use it. We developed a web application, which can change pitch on synthesised speech. We documented how adjusting the pitch can influence generated speech.

• Keywords: speech synthesis, text-to-speech, web application, fastpitch, emotions in speech