Beyond the ChatGPT composer
ChatGPT Dictation stays inside ChatGPT. WhisperJot types into whichever focused text field you are using, including other apps.
To dictate in ChatGPT, click the microphone icon in the composer, speak, and edit the text before sending. That Dictation control is separate from ChatGPT Voice. It works only inside ChatGPT and needs internet; WhisperJot instead types into whichever desktop app currently has focus.
ChatGPT Dictation uses the microphone icon in the composer: tap it, speak, and your words appear as text you can edit before sending. It is distinct from ChatGPT Voice, which uses the sound-wave icon for a two-way spoken conversation. Dictation is available in the browser and desktop apps since April 2025, with mobile availability earlier. ChatGPT sends the recorded audio to its models to be transcribed, so internet is required. Audio is retained as long as the chat exists and is deleted within 30 days after the chat is deleted. Consumer audio is used for training only if Include your audio recordings is on; business workspaces are not trained on by default. Dictation works only inside ChatGPT's own composer and has no global hotkey for other apps.
ChatGPT Dictation stays inside ChatGPT. WhisperJot types into whichever focused text field you are using, including other apps.
Local engines transcribe on-device by default and can run offline after a one-time model download (about 500 MB on Mac, around 600 MB on Windows and Linux). Jot Cloud is opt-in.
Filler removal, spoken punctuation, self-correction, custom vocabulary, and spoken shortcuts shape the text before it is inserted.
Keep a prompt-friendly tone for ChatGPT while other apps use their own per-app writing behavior.
Open ChatGPT in the browser or desktop app, click in the composer, and tap the microphone icon in the text box. Speak, then edit the transcribed words before sending. Dictation has been available in the browser and desktop apps since April 2025, with mobile earlier. You need an internet connection because the recorded audio is sent to ChatGPT's models to be transcribed.
Dictation uses the microphone icon and turns your speech into editable text in the composer. ChatGPT Voice uses the sound-wave icon for a two-way spoken conversation rather than a text draft you revise before sending. If you want words you can edit, choose Dictation. If you want to talk back and forth by voice, that is the separate Voice feature.
ChatGPT sends recorded audio to its models to be transcribed, so the feature needs internet. Audio is kept as long as the chat exists and is deleted within 30 days after you delete the chat. On consumer accounts, audio is used for training only if Include your audio recordings is turned on. Business workspaces are not trained on by default.
No. ChatGPT Dictation works only inside ChatGPT's own composer and does not provide a global hotkey for other applications. To dictate into email, notes, a browser field, or another editor, use operating-system dictation in that field or a system-wide tool such as WhisperJot, which inserts text wherever the cursor is.
Use ChatGPT Dictation when you want the built-in microphone in the composer and are comfortable with its cloud transcription. Use WhisperJot when you want one hotkey that also works outside ChatGPT, local engines that run on-device by default, filler removal, custom vocabulary, or per-app tone. Jot Cloud is opt-in and is not required for the local engines.
Sources, checked 2026-09-01: ChatGPT voice dictation FAQ, ChatGPT Voice help, Using voice in ChatGPT. Vendor features change — check each product's documentation for current behavior.
Browse WhisperJot use cases, compare dictation apps, or check the system requirements.
One $99/year plan includes unlimited dictation across your focused desktop apps.