To test a free Mandarin transcription tool, send a short clip that has numbers and names, then count the wrong characters. Potok cites a published character error rate of 7.8 for Chinese with large-v3, from other people's tests, while the earlier model scored 14.7.
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
- The 7.8 is a character error rate for large-v3 in a published test, not our measurement.
- The earlier large-v2 model scored 14.7 in published tests, so the model version matters.
- Chinese is written without spaces, so errors are counted per character.
- Check whether the output uses the script form you expect.
What a transcript looks like
[00:00] Speaker 1: 我们开始开会吧。首先回顾上周的结果，然后规划下一步。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 |
|---|---|---|---|
| Chinese (Mandarin) (中文（普通话）) | Recommended | character error rate 7.8% (large-v3, FLEURS) Whale paper, Whisper paper | Published tests show a low error rate for this language. Chinese is written without spaces, so quality is reported as a character error rate. The earlier large-v2 model scored 14.7. |
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
- Prepare a clip with traps. Choose a minute of Mandarin that mixes numbers, company names and a person's name.
- Open the bot. Open @potok_now_bot in Telegram and send the clip as audio, a voice message or a video.
- Look at the script. Chinese can appear in simplified or traditional characters. Note which one you received and whether it suits you.
- Count characters. With no spaces between words, mark each wrong or missing character and divide by the total you heard.
- Go long when it holds. A recording up to 2 GB is accepted. Use Upload in the mini app for big files and export TXT, DOCX or SRT.
| 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 |
|---|---|---|
| Expecting spaces between words | Chinese text has none, so word-based checks fail. | Judge by characters. |
| Ignoring numerals | Numbers can be written as digits or characters, and a mismatch causes confusion. | Decide how you want them written and check the sample. |
| Quoting the large-v2 score as the current one | The 14.7 belongs to the earlier model. | Cite 7.8 for large-v3 and name the model. |
| Mixing dialect speech into the test | Regional speech is not what the figure describes. | Use standard Mandarin for the first test. |
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
What is the published quality for Chinese?
A character error rate of 7.8 for large-v3, and 14.7 for the earlier large-v2, both from published tests.
Why characters and not words?
Chinese is written without spaces, so a word count is not reliable and characters are the unit.
Will I get simplified or traditional characters?
It can vary, so check a short sample. The transcript is not converted for you afterwards.
Can I check a doubtful line later?
Yes. Turn timestamps on and jump to that moment in the audio.
Does the bot render the text in English?
No. The transcript stays in the language of the recording.
Sources
Each source was opened by hand on the date shown. Pages change, so open the source if a detail matters to you.
- Whisper speech recognition system (Wikipedia), Wikipedia · checked
- Word error rate (Wikipedia), Wikipedia · checked
- Whale paper, Table 5 (Whisper large-v3 on FLEURS) · checked
- Whisper paper, Table 13 (large-v2 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