Free Portuguese transcription is best judged with two clips, one Brazilian and one European. Published tests of others give large-v3, the model behind Potok, a word error rate of 3.65 for Portuguese in general. Read each sample against its audio and note where the accents differ.
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
- That score covers Portuguese as a whole; Brazilian and European accents were not reported separately in the tests.
- Spelling also differs between the two main varieties, so some mismatches are style rather than hearing.
- Nasal vowels and unstressed endings are quiet and easily lost, which makes them a useful probe.
- The figure belongs to other people's benchmark recordings, not to anything we measured.
What a transcript looks like
[00:00] Speaker 1: Vamos começar a reunião. Primeiro, vamos revisar os resultados da semana passada e depois planejar os próximos passos.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 |
|---|---|---|---|
| Portuguese (Português) | Recommended | word error rate 3.65% (large-v3, FLEURS) Canary-1B-v2 paper | Published tests show a low error rate for this language. The published figure is for Portuguese in general; Brazilian and European accents were not reported separately. |
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
- Get a pair of clips. Find about forty seconds from a Brazilian speaker and the same length from a European one, if you work with both.
- Open the bot. Open @potok_now_bot in Telegram and send each clip as a voice message, an audio file or a video.
- Check nasal endings. Words ending in nasal sounds are where the text most often goes soft. Compare them with what you hear.
- Note the spelling variety. Decide whether the output matches the variety your readers expect, and record any difference.
- Run the longer file. If both clips read well, send the full recording, with large files up to 2 GB going through the Upload button, 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 |
|---|---|---|
| Treating one accent as proof for all | A tool can be strong on one variety and slip on the other. | Test with the accent you will really receive. |
| Marking spelling variety as a hearing error | Your tally rises without any real failure. | Count it as a style note. |
| Ignoring unstressed endings | Dropped endings change tense and number. | Read the last syllable of each verb. |
| Using a clip with heavy background music | Instrumentals cover quiet vowels. | Pick a stretch of plain speech. |
More
See the full list of languages, the tools and the glossary.
Frequently asked questions
Does the Portuguese score cover Brazil and Portugal?
It is for Portuguese in general, because the published tests did not separate the accents.
Is 3.65 your own measurement?
No. It comes from published tests of others on large-v3.
Will the text follow Brazilian spelling?
It may vary, so check a sample and fix a few words by hand if your readers expect one variety.
Can the transcript be switched to English?
No. The transcript stays in the language of the recording.
How large a file can I use?
Files up to 2 GB are accepted, and a long result arrives as a file.
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