Hungarian speech to text with Potok is a borderline case. Published tests of large-v3 report a word error rate of 14.11, close to the 15 percent line we use, taken from other people's work. Run a short sample before you send a long recording.
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
- 14.11 sits just under the line we use, so a small difference in your audio could push it over.
- Hungarian attaches many endings to a word, and a word-based score counts a wrong ending as a whole wrong word.
- Only one published source was found for this language.
- Names often appear with the family name first, which is worth remembering when you check the text.
What a transcript looks like
[00:00] Speaker 1: Kezdjük el az értekezletet. Először átnézzük a múlt heti eredményeket, aztán megtervezzük a következő lépéseket.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 |
|---|---|---|---|
| Hungarian (Magyar) | Recommended, borderline | word error rate 14.11% (large-v3, FLEURS) Canary-1B-v2 paper | Published tests show an error rate close to our 15% line. Check a short sample first. This figure is close to the 15% line we use, so treat it as borderline. |
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
- Run the sample first. Borderline means a short test costs little and tells you a lot. Send one minute of the real speaker.
- Use the bot. Open @potok_now_bot in Telegram and attach the clip.
- Read for endings. Check suffixes on nouns and verbs, where slips are most likely.
- Send the full file. If the sample is acceptable, send up to 2 GB, using Upload for big files.
- Export. TXT, MD, DOCX, PDF, SRT or VTT.
- Keep the original. Store the source audio so any doubtful suffix can be replayed.
| 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 |
|---|---|---|
| Sending a long file without a test | A borderline language can surprise you. | Always run a sample. |
| Treating 14.11 as safe | It is near the cut-off. | Plan to proofread. |
| Judging only by meaning | A wrong ending changes grammar. | Read closely. |
| Calling it our measurement | It is published by others. | Cite it that way. |
| Searching for a word stem in the output | Endings change the written form, so the exact word may not match. | Search for a shorter stem. |
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
Why is Hungarian borderline?
The published word error rate is 14.11, close to the 15 percent line we use.
Is the figure our own?
No, it comes from published tests by others and from a single source.
How do I test a clip?
Send a short sample to the bot and compare the text with the audio before sending the full file.
Can I send long recordings?
Yes, up to 2 GB per file, with Upload for the biggest.
Does the text stay Hungarian?
Yes. The bot does not translate, so it stays in Hungarian.
Should I proofread more than for other languages?
Yes. For a borderline language plan to read the whole text against the audio before you share it.
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