To transcribe an interview, send the recording to the Potok bot on Telegram, switch on speaker labels and timestamps, and check every quote against the audio. Files up to 2 GB work. Export DOCX to edit, or TXT, MD, PDF, SRT and VTT. The bot works in about 30 languages.
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
- Treat the transcript as a draft: names, numbers and overlapping voices are where mistakes cluster.
- Speaker labels are numbers such as Speaker 1. Rename them to real names in the exported file.
- Timestamps take you from a quote straight back to its place in the audio.
- Ask for consent before you record, and say how you will use the text.
The routine in six steps
An interview is the recording where text pays off most: you need the exact words, who said them, and a way back to the right minute.
- Get consent. Tell the person you are recording and how you will use the recording and the text. Rules differ by country, so check the ones that apply to you.
- Record close to the voices. Use a quiet room and put the microphone near the speakers, not in the middle of the table.
- Say the hard names aloud. At the start, say the names of people, companies and places, and spell the unusual ones.
- Send the recording. Open @potok_now_bot in Telegram and send the audio or video. For a large file, up to 2 GB, use the Upload button of the bot's mini app.
- Switch on speaker labels and timestamps. Do it before you send the file, so every passage starts with a time and a speaker number.
- Clean the text with the audio playing. Rename the speakers, then fix names, numbers and the places where two people spoke at once.
| 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 |
What the transcript looks like
With labels and timestamps on, each passage starts with a time and a speaker number. The example below is text we wrote to show the layout. It is not a real interview.
[00:00] Speaker 1: Thank you for meeting me. Let's start with the budget.
[00:04] Speaker 2: Of course. The plan changed in the spring, so I'll begin there.
[00:10] Speaker 1: And who approved the change?
[00:13] Speaker 2: The board did, at its last meeting before the summer.
Sample text written by us to show the layout of a transcript. It is not the output of the recognition engine and not a real recording.
Clean the transcript
Replace the numbers with names in your editor. Then read the text with the audio playing and fix what you hear. Give most attention to names of people, companies and places, to numbers and amounts, and to places where two people spoke at once. The labels come from speaker diarization, which separates voices by sound and cannot know who the people are.
Quote with care
Before you publish a quote, listen to it once more. If the interview is in a language other than English, ask a speaker of that language to read the quotes you use. How well the languages are covered is on the list of languages.
Choose the file
DOCX opens in Word and Google Docs and is easy to comment on. TXT and MD are best for notes. SRT and VTT are subtitle files, handy if the interview is on video. Starting from an MP3? See MP3 to text. Recording from a messenger? See Telegram voice messages.
Common mistakes
| Mistake | What happens | Fix |
|---|---|---|
| Putting the microphone in the middle of the table | Both voices arrive weak and overlapping, and the labels start to mix up. | Move the microphone close to the speakers or give each person a microphone. |
| Quoting straight from the transcript | A wrong word in a quote changes what the person said. | Listen to every quote once more before you publish it. |
| Leaving Speaker 1 and Speaker 2 in the final text | Readers cannot tell who is speaking, and the labels are not real names. | Replace the labels with names in the DOCX or TXT. |
| Skipping the spelling of names | Rare names come out as the nearest common word. | Say and spell hard names at the start, and keep a list to check against. |
Frequently asked questions
Should I ask permission before I record an interview?
Yes. Ask the person before you record, and tell them how you will use the text. Rules differ by country and by situation, so check the rules that apply to you.
Can the transcript tell who is speaking?
Speaker labels can be switched on. They are numbers such as Speaker 1 and Speaker 2. You replace them with names in the exported file.
Is the transcript good enough to quote?
Treat it as a draft. Names, numbers and overlapping voices are where mistakes happen. Check every quote against the audio before you publish it.
Which export format is best for an interview?
DOCX is easy to edit and comment on. TXT or MD is best for notes. SRT or VTT helps if you add subtitles to a video interview.
Can I transcribe an interview in another language?
Yes, in about 30 languages with good published results and a few more with warnings. See the full list for the verdict on your language.
Sources
Each source was opened by hand on the date shown. Pages change, so open the source if a detail matters to you.
- Speaker diarisation (Wikipedia), Wikipedia · checked
- SubRip and the .srt format (Wikipedia), Wikipedia · 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