Apps

Voice input for ChatGPT: voice mode and text dictation

Mark Vesyoly · founder of Potok · ~5 min

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

The ChatGPT apps have their own voice input, but it is a voice mode for conversation, not line-by-line dictation into the prompt field. To get exact text in the field, dictated word for word, you need a third-party tool: a built-in OS tool or Potok (potok.now), which inserts dictated text into the active field and recognizes about 900 characters per minute.

Want to dictate a long prompt into ChatGPT instead of typing it by hand? The ChatGPT apps have their own voice input, but it is built differently from dictation into a text field, and it is worth understanding this before you start speaking a request.

What voice input in ChatGPT itself can do

The ChatGPT apps have their own voice input, which the company calls the voice mode. The main thing known about it is that it is not classic line-by-line dictation, where what you say turns word for word into text in the input field. The voice mode works differently and does not replace dictating a specific prompt text.

What this means in practice: if you need to say a request in exact words and get ready text in the input field, the built-in voice input is not meant for that. The prompt text field itself does accept ordinary pasting of text, so you can paste text dictated with any third-party tool, whether a built-in operating system tool or a separate app. The situation with voice input in Cursor is similar: a voice of its own for the chat and, separately, a text field where ready text is pasted.

What changes with Potok

Voice input for ChatGPT with Potok (potok.now) works not with the ChatGPT app specifically but with any active text field, including the prompt field in a browser or in an app. It works like this: you hold the hotkey (Ctrl+Space by default), say the prompt, release the key, and the text appears in the input field. Potok places punctuation and capital letters itself: you do not need to dictate “zapyataya” (comma) or “tochka” (period) separately.

Recognition runs in the cloud on Potok's own model trained on Russian speech, including terms and Russian text mixed with English words and commands, which is handy for prompts where some words are in English. By the stated figures, Potok recognizes about 900 characters per minute by voice, noticeably faster than ordinary typing (about 200 characters per minute). After that you can edit the ready text and send it as an ordinary prompt; dictating it in full is described on the page about voice input for prompts.

Platforms
Windows, macOS, Linux
Free
Free: 2500 words per week; full access for the first 14 days, no card needed
Pro
Pro 690 ₽/mo, 499 ₽/mo when paid yearly
Payment
Russian bank cards, rubles
Where speech is recognized
cloud, servers in Russia; internet required

Potok is a cloud service: speech is recognized on servers in Russia, and it does not work without the internet. Download: potok.now/download

When the built-in ChatGPT voice mode is enough

If you just need to talk to ChatGPT by voice and hear the answer, the built-in voice mode is enough, and you do not need a separate app for that. The app is needed when it is important to get a specific, precisely worded prompt text in the input field, dictated word for word.

Potok is a cloud service: recognition runs on servers in Russia, and it does not work without an internet connection. Potok has no offline mode, voice commands or phone app: it is a desktop program, not a replacement for the voice mode inside ChatGPT itself.

Frequently asked questions

Does ChatGPT dictate prompt text by voice the way ordinary dictation does?

No. The ChatGPT voice mode is a separate way of talking by voice, not line-by-line dictation into a text field. To get exact text in the prompt field, dictated word for word, you need a third-party tool: a built-in OS tool or a separate app such as Potok.

Can I write code and commit messages by voice?

By voice it is convenient to dictate text parts: comments, commit descriptions, prompts for an AI assistant, code review text. Potok has no direct voice control of code (commands to edit code by voice); this is not part of the product's functionality.

Does voice input understand mixed Russian and English speech when dictating code or prompts?

It depends on the recognition model: not all systems switch between languages inside one phrase (code-switching) equally well. The Potok model is trained on Russian speech, including terms and Russian mixed with English.

How do I dictate a prompt for ChatGPT in Russian with English terms?

You can dictate the whole prompt by voice: specialized services such as Potok are built for mixed speech (Russian text with English terms is exactly the scenario of vibe coding and working with AI), and then the text is inserted straight into the chat input field.

Voice input for neural networks: why do developers need it?

Developers often write long prompts for AI in Russian mixed with English terms (vibe coding). Dictating such a prompt by voice is usually faster than typing it, especially if the prompt is large.

Does voice input work in Cursor and other IDEs?

Potok works on top of the active text field in any app, including IDEs such as Cursor and VS Code: it is not tied to a specific program and inserts text where the cursor is.

Download Potok

Works with a single key in any program. 14 days of full access, then Free with 2500 words per week or Pro at 690 ₽/mo (499 ₽/mo paid yearly). Potok is a cloud service: speech is recognized on servers in Russia, and it does not work without the internet.

Download Potok

Existing recordings are transcribed by Potok's Telegram bot @potok_now_bot: send a file or a YouTube, VK Video or RuTube link and the text arrives in the chat. The Potok subscription removes the limit in the bot too.

Built-in dictation and Potok

Based on Microsoft and Apple help pages.
ToolWhere it worksRussian languageWhere speech is recognized
Windows voice typing (Win+H)Windows 10 and 11Windows 11 onlyMicrosoft cloud, does not work offline
macOS DictationmacOSyeson the device or on Apple servers
PotokWindows, macOS, Linuxprimary languagecloud, servers in Russia, does not work offline
Calculated from average typing and speaking speeds, not measured.
TaskCharactersTyping (200 chars/min)Voice (900 chars/min)
chat reply3002 min0.5 min
email15008 min1.7 min
post or meeting minutes400020 min4.4 min
article1000050 min11.1 min

Primary sources: chatgpt.com · help.openai.com

Mark Vesyoly — founder of Potok. Writes about Russian voice input: how to dictate in work apps, what gets in the way and how to fix it.
© Potok · Vladimir Olegovich Lipatnikov · INN 502915859701 · Public offer Comparisons · FAQ · About Free voice input · Voice input for AI chats · Data and Russian privacy law · Voice input on Mac · Bot plans · 14 days of Pro for a review