English speech to text with Potok is a recommended language with the lowest published error on our list. Large-v3 results run from 4.0 to 4.25 word error rate in published tests, and a third party averaged 4.80 for the turbo variant. These are others' tests, not our measurements.
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
- Three published figures sit between 4.0 and 4.80, so English is the best documented language here.
- Test sets of read speech are cleaner than real meetings, so expect more errors on your files than on the benchmarks.
- English accents vary widely, and a heavy regional accent or fast crosstalk is where the gaps show.
- Homophones are spelled by context, so a mix-up like their and there can still happen.
What a transcript looks like
[00:00] Speaker 1: Let's start the meeting. First, we'll review last week's results, then plan the next steps.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 |
|---|---|---|---|
| English | Recommended | word error rate 4% (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 4.0 to 4.25; the turbo variant averages 4.80 in a third-party test. |
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
- Start from the source. Phone voice notes, Zoom exports and lecture recordings are all fine. Keep the original file rather than a re-compressed copy.
- Send. Open @potok_now_bot and share the file or a link.
- Look for context errors. Scan for homophones and for proper nouns, which no score captures.
- Large recordings. Anything up to 2 GB; use the Upload button for the biggest.
- Choose an export. TXT, MD, DOCX, PDF, SRT or VTT. Short results show in the chat.
| 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 benchmark numbers on a noisy call | Clean test speech is easier than real rooms. | Treat the figures as a floor, not an average. |
| Not reading homophones | Spelling errors that sound right slip through. | Proofread with the audio. |
| Quoting 4.0 as our accuracy | It is a published figure from others. | Cite it with its source type. |
| Skipping speaker labels in a panel | Names of voices are not known, only numbers. | Turn labels on and rename later. |
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
What are the published scores?
Large-v3 ranges from 4.0 to 4.25 word error rate, and the turbo variant averages 4.80 in a third-party test.
Does it handle accents?
The published figures do not split by accent, so test a clip from the speaker.
Is English the best-supported language?
It has the lowest published error on our list, but real audio matters more than the number.
Can I send a YouTube link?
Yes, links to YouTube are accepted along with files up to 2 GB.
Do I get timestamps?
Yes, optionally, and they export to SRT and VTT.
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