What Slovenian transcription gets right and wrong
Slovenian is a smaller language for speech recognition, so the result is a usable draft that needs review. Standard speech from one clear speaker, such as a lecture, a presentation or a radio interview, comes out best. Check every name, place and technical term.
Dialects are the hard part. Slovenia has an unusually large number of them for its size, and speech from Prekmurje, Carinthia, the Littoral or rural Styria can be far from the written standard. The model writes what it hears as standard Slovenian, so dialect recordings need careful correction and sometimes rewriting.
Casual Ljubljana speech also drops vowels and shortens words. Expect the model to fill in standard forms or mishear shortened ones in informal conversations.
For interviews and research, a practical approach is to fix the passages you plan to quote first, then work through the rest using the timestamps. If a speaker uses strong dialect throughout, decide early whether you want a standard Slovenian transcript or a dialect one, because the draft will lean toward the standard and you will be correcting in one direction.
Slovenian spelling and grammar to check
Slovenian uses the Latin alphabet with č, š and ž. These are the common fixes.
- Č, š and ž may become c, s and z, especially in names and fast speech.
- The dual (midva, gresta, dve knjigi) is unique to Slovenian among the languages around it, and the model may write a plural instead.
- Final l is often pronounced like u or w, so words like videl or bil can be misheard.
- Foreign names may be adapted to Slovenian spelling or kept in the original, and case endings may be added after a hyphen or not.
- Numbers may be written as digits or words. Slovenian uses a comma as the decimal mark and a period after day and month in dates.
Slovenian, Croatian and auto-detect
Slovenian is a South Slavic language, close enough to Croatian and Serbian that auto-detect can mix them up on short clips. A transcript that comes back with Croatian spellings, or in Serbian Cyrillic, usually means the language was misjudged. Clips that start in English or German can also be labeled wrongly.
This page preselects Slovenian. If the language is wrong, check the setting and run the file again. In mixed families or border areas where people switch between Slovenian and Croatian, Italian or German, choose the language spoken most in each file.
Short clips are the most likely to be misjudged, because a few seconds of speech give auto-detect little to go on. A WhatsApp voice note that opens with "Živjo" or a greeting shared with Croatian can easily be labeled as the wrong language, which is another reason to keep Slovenian selected before you upload.
English mixing and who uploads
Younger speakers and tech teams mix in English words. They usually come through in English spelling, sometimes with Slovenian endings. If a meeting is mostly English, choose English.
Common uploads are university lectures from Ljubljana and Maribor, research interviews, meetings, local podcasts, radio programs and YouTube videos that need subtitles. Slovenian communities in Austrian Carinthia, around Trieste in Italy, in Argentina and in the US (Cleveland especially) upload family recordings, and oral history projects record older speakers whose dialect needs extra attention.
Uploading and exports
Upload MP3, M4A, WAV, WhatsApp voice notes or MP4 and MOV video as they are, record in the browser, or import a YouTube or TikTok link. The transcript arrives in timestamped paragraphs. Click a line to hear it and edit the text in place.
Exports to TXT, DOCX, PDF, SRT, VTT and CSV keep the Slovenian letters. To read the content in English, use the Google Translate link or the AI prompt export with ChatGPT, Claude or Gemini. Speaker labels are on paid plans or with a Minute Pack. Files are never used to train AI models.