Czech works well in Potok: send speech to the Telegram bot and get Czech text back. Published tests of the large-v3 model give word error rates from 11 to 11.33, close enough to call one result. Voice messages, audio, video and files up to 2 GB are accepted.
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
- Published large-v3 word error rates for Czech range from 11 to 11.33, depending on the test.
- The numbers come from published papers and public pages, not from tests we ran.
- Czech marks such as the caron in č, š, ž, ř and ě, and the acute accent on long vowels, change words.
- Names are declined in Czech, so check them against the base form.
What a transcript looks like
[00:00] Speaker 1: Začněme schůzku. Nejdříve projdeme výsledky z minulého týdne a potom naplánujeme další kroky.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 |
|---|---|---|---|
| Czech (Čeština) | Recommended | word error rate 11% (large-v3, FLEURS) Canary-1B-v2 paper, Whale paper | Published tests show a low error rate for this language. Published large-v3 results range from 11.0 to 11.33. |
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
- Open the bot. Find @potok_now_bot and forward a Czech voice message or send audio, video or a YouTube link.
- Watch the file size. The limit is 2 GB; larger lectures can be loaded with the Upload button of the mini app.
- Add labels for panels. Use speaker labels and timestamps for a debate or meeting, so you can see where each speaker changes.
- Search for the tricky letters. Look for ř, ě, ů and long vowels, since a missing mark can produce a different word.
- Export. Choose the format that matches your next step: TXT, MD, DOCX, PDF, SRT or VTT.
| 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 |
|---|---|---|
| Reading the 11 as a promise for your own file | Noisy or overlapping speech can be much worse. | Treat it as a general level seen in tests, not as a result for your audio. |
| Skipping a check of long and short vowels | Words that differ only by a length mark get confused. | Proofread the key sentences. |
| Pasting a declined surname into a database | The same person ends up listed twice. | Return the name to the nominative form. |
| Waiting for a translation | The transcript remains Czech. | Translate the final text elsewhere. |
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
What error rate is published for Czech?
Large-v3 results range from 11 to 11.33 in published tests. They are other groups' numbers and not measurements of your audio.
Why do two figures exist for Czech?
Different papers use different test sets and text rules, so the same model gives slightly different numbers.
Does it write the háček and the čárka?
It aims to, but check them. Missing marks are a typical slip and can change the meaning of a word.
Can I transcribe a Czech YouTube video?
Yes. Paste the link into the chat, within the limits for file size.
Is there a translated version?
No. The text stays in the original language.
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
- Whisper speech recognition system (Wikipedia), Wikipedia · checked
- Canary-1B-v2 paper, Table 11 (Whisper large-v3 on FLEURS) · checked
- Whale paper, Table 5 (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