Details

Podporna orodja za raziskovanje govorjene slovenščine
ID Križaj, Janez, Univerza v Ljubljani, Fakulteta za elektrotehniko (Author), ID Dobrišek, Simon, Univerza v Ljubljani, Fakulteta za elektrotehniko (Author)

.pdfPDF - Presentation file, Download (859,80 KB)
MD5: C5C3D1650CD8F2CA3099377516A158AB
URLURL - Source URL, Visit https://journals.uni-lj.si/slovenscina2/article/view/22946 This link opens in a new window

Abstract
Članek predstavlja pet odprtokodnih orodij za obdelavo govornih posnetkov, razvitih za podporo raziskovanju govorjenega slovenskega jezika. Orodja pokrivajo različne vidike dela z govornimi jezikovnimi podatki in se med seboj dopolnjujejo. Prvo orodje omogoča vsiljeno poravnavo govora z besedilom s pomočjo orodij programske knjižnice Montreal Forced Aligner. Drugo orodje je namenjeno anonimizaciji zvočnih posnetkov, kjer sistem samodejno prepoznane ali ročno označene dele posnetkov z osebnimi podatki nadomesti z zvočnimi maskami. Tretje orodje služi validaciji govornih posnetkov in z grafičnim vmesnikom omogoča preverjanje ujemanja posnetka s transkripcijo. Četrto orodje je akustični normalizator, ki združuje različne metode za odstranjevanje šuma in izboljšanje kakovosti posnetkov. Peto orodje pa je grafemsko-fonemski pretvornik, ki samodejno pretvarja besedila iz običajnega grafemskega zapisa v fonetično transkripcijo.

Language:Slovenian
Keywords:govorni posnetki, vsiljena poravnava, anonimizacija, validacija, akustična normalizacija, grafemsko-fonemska pretvorba
Work type:Article
Typology:1.01 - Original Scientific Article
Organization:FE - Faculty of Electrical Engineering
FF - Faculty of Arts
Publication status:Published
Publication version:Version of Record
Publication date:09.10.2026
Year:2026
Number of pages:Str. 88-106
Numbering:Letn. 14, št. 1
PID:20.500.12556/RUL-190071 This link opens in a new window
UDC:004.934:811.163.6
ISSN on article:2335-2736
DOI:10.4312/slo2.0.2026.1.88-106 This link opens in a new window
COBISS.SI-ID:301006851 This link opens in a new window
Publication date in RUL:21.10.2026
Views:99
Downloads:21
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Record is a part of a journal

Title:Slovenščina 2.0 : empirične, aplikativne in interdisciplinarne raziskave
Publisher:Trojina, zavod za uporabno slovenistiko, Trojina, zavod za uporabno slovenistiko, Trojina, zavod za uporabno slovenistiko, Znanstvena založba Filozofske fakultete, Znanstvena založba Filozofske fakultete, Založba Univerze v Ljubljani
ISSN:2335-2736
COBISS.SI-ID:264547328 This link opens in a new window

Licences

License:CC BY-SA 4.0, Creative Commons Attribution-ShareAlike 4.0 International
Link:http://creativecommons.org/licenses/by-sa/4.0/
Description:This Creative Commons license is very similar to the regular Attribution license, but requires the release of all derivative works under this same license.

Secondary language

Language:English
Title:Support tools for spoken Slovene research
Abstract:
The article presents five open-source tools for processing speech recordings, developed to support research on spoken Slovene. The tools cover different aspects of working with spoken language data and complement each other. The first tool enables forced alignment of speech with text using the Montreal Forced Aligner software library. The second tool is intended for anonymization of audio recordings, where the system replaces automatically detected or manually marked segments containing personal data with audio masks. The third tool is used for validation of speech recordings and, through a graphical interface, allows verification of the correspondence between the recording and its transcription. The fourth tool is an acoustic normalizer, which integrates various methods for removing noise and improving the quality of recordings. The fifth tool is a grapheme-to-phoneme converter that automatically converts text from standard graphemic notation into a phonetic transcription.

Keywords:speech recordings, forced alignment, anonymization, audio validation, acoustic normalization, grapheme-to-phoneme conversion

Projects

Funder:ARIS - Slovenian Research and Innovation Agency
Project number:J7-4642
Name:Temeljne raziskave za razvoj govornih virov in tehnologij za slovenščino

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Collection

This document is a part of these collections:
  1. Slovenščina 2.0

Back