Web, Windows, and other apps
Quick Entry dictation is documented for Claude Desktop on Mac. WhisperJot types into claude.ai, other Claude surfaces you can focus, and the rest of the desktop.
On Claude Desktop for Mac, enable Quick Entry dictation in Settings > General, then use Caps Lock to start and stop speech-to-text. The claude.ai web composer has no documented dictation button, so use OS dictation or WhisperJot there. Anthropic's Voice mode is a separate two-way conversation feature.
Anthropic separates Voice mode from dictation: Voice is a two-way conversation via the sound-wave button (beta on mobile, desktop, and web), while dictation converts speech to text so you can type prompts. Claude Desktop for Mac includes Quick Entry dictation: enable it in Settings > General, where it takes over Caps Lock and needs macOS 14 or later plus speech-recognition permission; press Caps Lock to start, press it again to stop, then click the arrow to send. The claude.ai web composer has no documented dictation button—use OS dictation there. Windows desktop dictation is not documented. Where Anthropic documents processing (Claude for Microsoft 365 and Claude Code), audio is streamed to Anthropic's servers for cloud transcription and is not retained afterward; mobile dictation deletes the recording after converting it and does not use it for training.
Quick Entry dictation is documented for Claude Desktop on Mac. WhisperJot types into claude.ai, other Claude surfaces you can focus, and the rest of the desktop.
WhisperJot's local engines transcribe on-device by default and work fully 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, and custom vocabulary help spoken prompts read more like typed ones.
Keep a prompt style for Claude while chat apps or email use their own per-app tone settings.
Open Claude Desktop for Mac, go to Settings > General, and enable Quick Entry dictation. It takes over Caps Lock and requires macOS 14 or later plus speech-recognition permission. Press Caps Lock to start, speak your prompt, press Caps Lock again to stop, then click the arrow to send. Anthropic treats this as dictation—speech converted to text—not the separate Voice mode conversation.
Anthropic has not documented a dictation button in the claude.ai web composer. Voice mode, which uses the sound-wave button for two-way conversation, is a different feature and is in beta on mobile, desktop, and web. To type a prompt by voice on the website, use macOS Dictation or Windows voice typing in the focused field, or a system-wide tool such as WhisperJot.
Anthropic says dictation converts your speech to text so you can type prompts. Voice mode is a two-way spoken conversation started from the sound-wave button and is described as beta on mobile, desktop, and web. If you want editable prompt text, use Quick Entry dictation on Mac, OS dictation in the web composer, or WhisperJot. Do not expect Voice mode to behave like a compose-field dictation control.
Treat Anthropic dictation as cloud transcription. Where Anthropic documents processing—for Claude for Microsoft 365 and Claude Code—audio is streamed to Anthropic's servers and is not retained after transcription. On mobile, dictation deletes the recording after converting it and does not use it for training. This guide does not claim a processing location beyond those documented products.
Claude Windows desktop dictation is not documented, so this guide does not describe a built-in Windows control. WhisperJot still types into whichever Claude field has focus on Windows or Linux: click the prompt, press Ctrl+Alt+Space, speak, and the text is inserted. On the web, the same idea applies with OS dictation or WhisperJot because claude.ai has no documented composer dictation button.
Sources, checked 2026-09-01: Claude Voice mode help, Claude Desktop Quick Entry on Mac, Claude dictation documentation, Claude mobile dictation help. 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.