Potok converts Vietnamese audio to text when you send a recording to the Telegram bot. Published tests of others on the earlier large-v2 model, not our measurements, give a word error rate of 10.3, while another test set gave 18.9 for large-v3, so results vary with the data. Files up to 2 GB work.
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
- Two published figures disagree, 10.3 and 18.9, because the test sets and models differ.
- Vietnamese puts a space between every syllable, so a word error rate counts syllables more than whole words.
- Stacked tone marks and extra vowel signs must all survive the export.
- A single wrong tone mark gives a different word, so tones are the main thing to proofread.
- Names are written with a space between each syllable, so a surname and a given name need a manual check for order.
What a transcript looks like
[00:00] Speaker 1: Chúng ta bắt đầu cuộc họp nhé. Trước tiên, chúng ta sẽ xem lại kết quả tuần trước, sau đó lên kế hoạch cho các bước tiếp theo.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 |
|---|---|---|---|
| Vietnamese (Tiếng Việt) | Recommended | word error rate 10.3% (large-v2, FLEURS) Whisper paper | Published tests show a low error rate for this language. The figure is from the earlier large-v2 model; another test set gave 18.9 for large-v3, so results vary with the data. |
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
- Expect a wide range. Because published results spread from 10.3 to 18.9, test a clip from your own recording.
- Open the bot. Open @potok_now_bot and send a sample.
- Send the whole file. Up to 2 GB; Upload in the mini app handles the biggest.
- Check tone marks. Look at a paragraph and compare the marks with the audio.
- Check names. Personal and place names are easy to get wrong.
- Export. DOCX or TXT to edit, SRT or VTT for subtitles.
| 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 |
|---|---|---|
| Quoting only the low number | Another test gave 18.9. | Quote the spread. |
| Dropping tone marks in an old editor | Words change. | Use Unicode. |
| Counting syllables as words | The metric looks inflated. | Read the score as an indication. |
| Mixing northern and southern speech in one file | Regional forms confuse the text. | Review those parts. |
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
Which figure is right, 10.3 or 18.9?
Both are published. The first is from large-v2 and the second from large-v3 on another data set, so treat the spread as the answer.
Do tone marks survive?
Yes. The transcript is written in Vietnamese with its marks, so use Unicode software.
What is the size limit?
2 GB per file, with Upload for large ones.
Did you measure these?
No. They come from published tests by other people.
Is the text translated?
No. The transcript stays in Vietnamese, the same language as the recording you send, and it is never converted into another one.
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
- 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