Detecting artifacts in synthetic speech

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F49777513%3A23520%2F15%3A43926635" target="_blank" >RIV/49777513:23520/15:43926635 - isvavai.cz</a>
Result on the web
—
DOI - Digital Object Identifier
—

Result language
angličtina
Original language name
Detecting artifacts in synthetic speech
Original language description
Nowadays, speech synthesis is growing very popular in everyday use. For example, automatic voice assistants on mobile platforms are getting smarter every year, using speech synthesis and speech recognition to communicate with the user in a more natural way. As more people make use of speech synthesis, the quality requirements are higher more than ever. Although scientists currently focus mainly on HMM-based synthesis, real applications still use the traditional unit-selection method. Unit selection is known for its ability to produce high-quality synthetic speech. It produces more natural speech, but it may suffer from sudden quality drops at concatenation points. Quality drops ("artifacts") can theoretically occur at every concatenation point. In thefollowing paragraphs, an experiment on the automatic detection of artifacts in concatenation speech synthesis is presented. The main goal was to build a classifier which would mark suspicious segments in synthetic speech in the same way a
Czech name
—
Czech description
—

Project
<a href="/en/project/ED1.1.00%2F02.0090" target="_blank" >ED1.1.00/02.0090: NTIS - New Technologies for Information Society</a><br>
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>S - Specificky vyzkum na vysokych skolach

Publication year
2015
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Similar results(10)