Support

macOS — Download directly from our website and register with your email. No approval needed.

Download Flow Keyboard for macOS

iOS — Available free on the App Store. Download here.

Flow Keyboard is an AI voice input method for macOS and iOS. You speak naturally, and Flow converts your words into polished, well-structured text -- typically in under 2 seconds.

Unlike regular dictation, Flow intelligently refines your speech: removing filler words ("um", "uh", "like"), fixing grammar, adding punctuation, and organizing scattered thoughts into clear writing. It supports 30+ languages with automatic mixed-language detection, and can translate your speech into 16 target languages.

On macOS it works system-wide in any app; on iOS it works as a third-party keyboard. All data is discarded immediately after processing -- zero server-side retention. An internet connection is required.

Voice recognition (30+ languages) -- English, Chinese (including dialects like Cantonese), Japanese, Korean, French, German, Spanish, Portuguese, Italian, Russian, Thai, Vietnamese, Arabic, Indonesian, and more. Mixed-language input is automatically detected -- no manual switching needed.

Translation mode (16 target languages) -- Speak in one language, output in another. For example, speak Chinese and get English text, or speak English and get Japanese output. Targets include Chinese (Simplified & Traditional), Cantonese, English, Japanese, Korean, and 10+ European and Asian languages.

macOS

1. Download from flowkeyboard.com and install
2. Open Flow Keyboard — it appears as an icon in your menu bar
3. Grant Accessibility and Microphone permissions when prompted
4. Press and hold the Fn (Globe) key to record, release to finish

iOS

1. Download Flow Keyboard from the App Store
2. Go to Settings › General › Keyboard › Keyboards
3. Tap "Add New Keyboard" and select "Flow Keyboard"
4. Tap "Flow Keyboard" again and enable "Allow Full Access"
5. When typing in any app, tap and hold the globe icon to switch to Flow Keyboard

Yes. Flow maintains zero server-side data retention. Your audio is sent to servers for real-time processing, the polished text is returned, and all data (audio, transcriptions, text) is discarded immediately. Nothing is stored at any point.

See our Privacy Policy for full details.

Flow Keyboard offers a free trial with full access for the first month — no limits. After the trial, choose Plus (from $2.50/month) for 12,000 words/week, or Pro (from $8.33/month) for unlimited usage. Without a subscription, a free quota is still provided so you can continue using basic features. See our pricing page for details.

Apple Dictation transcribes speech word-for-word. Flow goes further with AI refinement:

• Removes filler words -- Apple Dictation keeps them
• Fixes grammar and restructures text -- Apple Dictation transcribes verbatim
• Auto-detects mixed languages -- Apple Dictation requires manual switching
• Translates into 16 languages -- Apple Dictation has no translation
• Up to 5 minutes recording -- Apple Dictation has a 60-second limit

Apple Dictation is free. Flow offers a free trial, then from $2.50/month (Plus). Flow maintains zero server-side data retention. See detailed comparison →

On macOS, Flow works as a system-wide input method in any app with a text field. On iOS, it works as a third-party keyboard. Common uses:

• Messaging -- Slack, iMessage, WhatsApp, Teams, Discord
• Email -- Gmail, Outlook, Apple Mail
• Documents -- Notion, Google Docs, Word, Obsidian
• Code -- Claude Code, Cursor, VS Code, ChatGPT

Under 2 seconds for short recordings (a few sentences). Longer recordings take proportionally more time, with support for up to 5 minutes. Speed depends on recording length, internet speed, and server load.

This is a known WeChat issue -- WeChat doesn't always detect newly installed third-party keyboards.

Solution: Fully close WeChat (swipe up from the app switcher to quit), then reopen it. After restarting WeChat, Flow Keyboard should appear and the keyboard switch button should be visible.

You're probably long-pressing instead of tapping. Flow on iOS uses tap-to-record:

• Tap once to start speaking
• Tap again to stop and process

Long-press opens the mode selector (Dictation or Translation). After choosing a mode, recording starts when you release. It does not work as "hold to speak."

For normal use, just tap, speak, and tap again.

Yes. Open Flow Keyboard Settings and click Shortcuts. You'll see a "Record" button -- click it and press your preferred key combination to replace the default shortcut.

You can also enable an alternate shortcut, which is useful when an external keyboard is connected.

This is usually a network issue. Try these checks:

1. Region settings -- Make sure your device's region matches your actual location.
2. VPN -- If a VPN is enabled, try disabling it and test again.
3. Wi-Fi -- Try disconnecting from Wi-Fi and using mobile data instead.

Flow requires an internet connection for processing, so network quality directly affects speed.

Flow actually works better with soft speech and whispers. You don't need to speak loudly -- even whispering into your phone in a quiet office is enough for accurate recognition.

Not happy with the output? Tap the magic wand button (bottom right) after text appears, then say what you want changed -- just like telling a person how to edit.

Examples of what you can say:

• "Fix the typo in the second sentence"
• "Add punctuation"
• "Make it more casual"
• "Remove the line breaks"
• "Add some emojis"

On iOS, hold the record button and select Translation mode. On macOS, press Fn+Shift to start translating and Fn to stop. Then speak in any language and Flow will output text in your chosen target language. For example, speak Chinese and get English text, or speak English and get Japanese output.

Select a word in the output and click "Add to Dictionary" to save it. You can also add words manually. This is especially useful for terms that are easily misrecognized:

• People's names
• Product names
• Brand names
• Uncommon or specialized terms

Adding these helps Flow recognize them accurately in future recordings.

Flow really shines with longer recordings -- 30 seconds to 5 minutes. When you speak a complete thought (an article draft, a journal entry, meeting notes), Flow organizes your logic clearly, structuring paragraphs and ideas in a way that short dictation tools can't match.

Don't hesitate to speak for a few minutes straight. The longer and more complete your thought, the better the output.

After generating output, you can use voice editing to revise it multiple times -- just keep tapping the magic wand and giving instructions. Once you get the hang of it, you can handle complex edits entirely by voice -- no manual typing needed. You can also undo and redo between different versions to compare results.