Croatian speech to text through Potok sits in the recommended group. Published tests of the large-v3 model give Croatian a word error rate of 11.07, taken from other people's work, not from our own checks. Send a voice message, recording or video to the Telegram bot.
Transcribe a recording
Send your voice message, audio or video to @potok_now_bot on Telegram and the text comes back in the chat.
Open @potok_now_bot in Telegram
Files up to 2 GB · export to TXT, MD, DOCX, PDF, SRT and VTT · speaker labels and timestamps · about 30 languages
Key facts
- 11.07 is one published figure, so keep a sceptical eye on long files with poor audio.
- Croatian spelling uses letters with marks such as c with a caron and d with a stroke, so check how names come out.
- The transcript stays in Croatian.
- Speaker labels and timestamps are optional.
What a transcript looks like
[00:00] Speaker 1: Počnimo sastanak. Prvo ćemo pregledati rezultate prošlog tjedna, a zatim isplanirati sljedeće korake.Meaning in English: Let's start the meeting. First, we'll review last week's results, then plan the next steps.
Written by us to show the script and the layout of a transcript. It is not the output of the recognition engine.
Published quality
| Language | Verdict | Published result | Notes |
|---|---|---|---|
| Croatian (Hrvatski) | Recommended | word error rate 11.07% (large-v3, FLEURS) Canary-1B-v2 paper | Published tests show a low error rate for this language. |
These numbers come from published tests by other people, not from tests by Potok. Your own audio, with its noise, accents and microphone, will differ.
How to do it
- Sort your audio. Group recordings by speaker count. A single voice is easiest.
- Open Telegram. Start @potok_now_bot and send the file.
- Check the letters. Look at words with special letters and at surnames; these show problems earliest.
- Send long ones the right way. Up to 2 GB through the Upload button of the mini app.
- Save the text. Choose TXT, MD, DOCX, PDF, SRT or VTT.
- Keep the original. Store the source audio next to the transcript so a doubtful line can be checked later.
| Feature | Details |
|---|---|
| Where it works | Telegram bot @potok_now_bot; large files go through the bot's mini app |
| File size | Up to 2 GB |
| Export | TXT, MD, DOCX, PDF, SRT, VTT |
| Speakers and time | Speaker labels and timestamps can be switched on |
| Languages | About 30 languages with good published results; a few more with warnings |
| Translation | The transcript is not translated into another language |
| Dictation | A separate Potok desktop app lets you dictate text into any program; it recognises Russian speech |
Common mistakes
| Mistake | What happens | Fix |
|---|---|---|
| Ignoring special letters | A missing mark can change how a name reads. | Proofread names against a source. |
| Using one setting for interviews and lectures | Labels help interviews, but lectures only need timestamps. | Switch labels on only when voices change. |
| Passing off 11.07 as our measurement | It was published by others. | Write that in your notes. |
| Sending a very quiet clip | Faint speech gets guessed. | Normalise volume before sending. |
| Assuming the language setting covers neighbouring languages | Serbian and Bosnian speech may read differently. | Test a clip in the exact language you have. |
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
How good is Croatian?
It is in the recommended group, with a published word error rate of 11.07 for large-v3.
Is that figure yours?
No. It comes from published tests by others, not from a measurement we made.
Does the bot know Serbian or Bosnian?
We have no figures for them on this page, so test a clip before relying on it.
Which formats can I download?
TXT, MD, DOCX, PDF, SRT and VTT are all available for every transcript.
Can it label who speaks?
Yes, as Speaker 1, Speaker 2 and so on, but never by real name.
What should I proofread first?
Surnames, place names and numbers, because a wrong letter there changes meaning and cannot be guessed from context.
Sources
Each source was opened by hand on the date shown. Pages change, so open the source if a detail matters to you.
- Word error rate (Wikipedia), Wikipedia · checked
- FLEURS speech dataset (Google, Hugging Face), Hugging Face · checked
- Canary-1B-v2 paper, Table 11 (Whisper large-v3 on FLEURS) · checked
Try it on your own recording
Send your voice message, audio or video to @potok_now_bot on Telegram and the text comes back in the chat.
Open @potok_now_bot in Telegram