How to convert audio to text with a bot in Telegram

Potok transcribes Russian speech; this site has been translated into English for convenience.

The Potok Telegram bot @potok_now_bot converts audio to text: forward a file or a voice message in private messages, and a short recording comes back as text, a long one as a file. The bot also has a mini app for files up to 2 GB. Recognition is free for 60 minutes per month; for fast dictation of your own text the Potok app fits.

Do you have an audio recording, such as an interview, a lecture or a voice recorder note, and no time to transcribe it by hand? The Potok Telegram bot @potok_now_bot transcribes a file you send into text without installing programs. The bot is cloud-based: recognition runs on servers in Russia, and it does not work without the internet.

How to convert audio to text with a bot in Telegram

Transcription takes a few steps, without entering card details and without installing apps. No registration is needed: the bot works right after the /start command.

1. Open the bot @potok_now_bot in Telegram and press Start, which is the /start command.
2. Send an audio file: in private messages you can send it as a document, forward a voice message or a video note from another chat, or paste a link to YouTube, VK Video or RuTube. In a group where the bot is added as an administrator, it accepts voice messages, video notes and audio up to 10 minutes long.
3. Wait for the reply. A short recording comes back as text right in the chat, a longer one as a separate file; the file format (TXT or DOCX) is chosen in the bot settings, /settings.
4. Under the finished text the buttons Summary, Tasks and deadlines, Ask and Other format appear: you can get a short retelling, a list of tasks by speaker or an answer to your own question about the content of the recording.
5. The bot adds timestamps [mm:ss] and speaker labels (Speaker 1, Speaker 2) if you turn them on beforehand in /settings; by default it sends plain text without them.

What you get in reply

The text the bot returns is already proofread: punctuation and capital letters are placed, obvious recognition errors are fixed, but the bot does not shorten or retell the text by default; that is a separate Summary button. What you can get next:

• the finished transcript, as text in the chat or a TXT/DOCX file, depending on the length of the recording;
• an export in other formats: Markdown, PDF, and also SRT and VTT subtitles if you need the recording for a video;
• a short summary with the Summary button and a list of tasks with the Tasks and deadlines button: who does what and by when;
• an answer to your own question about the content of the recording with the Ask button, without reading the whole text again;
• the saved recording in the My recordings section, which you can open again at any time.

The same is available in a browser, in the bot's mini app: sign in by email or through Telegram, drag a file of up to 2 GB into the window or paste a link to YouTube, VK Video or RuTube. Recordings and access are shared between the bot and the web version. If you need a broader overview of transcription services for comparison, there is a separate article, a comparison of audio-to-text transcription services.

When the bot will not fit

The bot will not fit if the recording is not in Russian: it recognizes only Russian speech, and there is no translation into another language. You will not learn the speakers' names: the bot labels lines only as Speaker 1, Speaker 2, without names. Transcription happens after the file or link is sent, not during the call itself: the bot is not built for a live broadcast during a conversation. It will not accept a link to Instagram or TikTok either: of the external sources the bot understands YouTube, VK Video and RuTube. And the bot cannot work without an internet connection: there is no offline transcription. Separately, why transcribing a conversation of several people is harder than dictating one voice is covered in the article why meeting transcription is harder than ordinary dictation.

Frequently asked questions

Primary sources: core.telegram.org · www.ixbt.com · developers.sber.ru · t.me