Call recording to text: transcription through the bot and in the browser

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

A call recording is transcribed to text through the Potok Telegram bot @potok_now_bot: forward the file or a voice message with the conversation recording, and the bot returns the ready text. For long calls and larger files the bot's mini app in the browser will do. 60 minutes of transcription a month are free; for dictating your own text by voice on a computer there is the separate Potok app.

A conversation by phone or in a video call sometimes has to be converted to text, for the minutes of a meeting, a quote in an article or a check of agreements. It usually comes down to two steps: find the recording file and get text from it, and sometimes the transcription of the call program itself is enough.

Where to get the call recording file

1. If the call is already saved as a file, whether a call recording on the phone, a voice recorder recording or a file from a video call program, find it in the device memory or in the history of the app where the conversation took place.
2. If there is no recording yet and you will need one, turn on voice or screen recording before the conversation starts: the exact way depends on the phone and on what you call with.
3. A voice message with a conversation recording sent in a messenger also works as the source file: it can be forwarded directly, without saving anything beforehand.
4. If the recording lies only in cloud storage, download the file to the device: transcription needs the file itself or a direct link to YouTube, VK Video or RuTube, not a link to a disk.

How to convert a call recording to text through the bot

1. Find the bot @potok_now_bot in Telegram and start it with the /start command.
2. Send the recording file (you can send it as a document) or forward a voice message with the conversation from another chat.
3. The bot sends the result: a short recording as text right in the chat, a long one as a separate file; the format for long replies (TXT or DOCX) is set in advance.
4. The buttons under the reply, “Summary”, “Tasks and deadlines”, “Ask”, “Other format”, give not the whole transcription but the part you need or an answer about the content of the recording.
5. It is worth turning on timecodes and speaker labels (/settings) in advance, before sending the file, if you will need them.
6. A heavier file is uploaded to the bot's mini app: sign-in by email or through Telegram, a limit of 2 GB per file; the result is equally visible there and in the bot itself.

The built-in transcription in a call program

Some video call programs and voice recorders on a phone can transcribe a conversation right during it or straight after; then the text is already ready and a separate service is not needed. If there is no such function, you do not want to pay for it, or the recording was made in advance, transcription by a separate service after the call will do, with any file, regardless of where the conversation took place. The difference between such instant transcription and ordinary voice input of text is covered in detail in the article how transcription differs from voice input.

When not to give the bot a call recording

Transcription in real time, right during the conversation, is not what the bot can do: it needs an already finished and saved file, and it does not accept a direct stream. Without the internet the Potok bot stops working: it is cloud-based, and the sound is recognized on servers in Russia. There is one recognition language, Russian; if the conversation was not in Russian or a ready translation is needed, the bot will not help. The bot will not tell you by name who was speaking in the recording: it tells participants apart only with the labels “Speaker 1”, “Speaker 2”. You can compare ready transcription services with each other in a separate analysis, the comparison of transcription services.

Frequently asked questions

Primary sources: kontur.ru · kontur.ru · audio-2-text.ru · transcribe-ai.ru