FnFlowHow 7 Mac dictation apps get you to your first dictation
A dictation app is no use until you've said your first sentence into it. Before that you usually get through an account, macOS permissions, a hotkey and often a model download. We compared how seven apps handle this and built FnFlow's setup around what worked best.
The comparison
| App | Account first | Permissions (order) | Input Monitoring | Default key | fn / 🌐 conflict | Speech model | Practice step |
|---|---|---|---|---|---|---|---|
| FnFlow 0.12 | No | Microphone → Accessibility (one button) | No | fn if 🌐 is free, else right ⌘ | Detects the 🌐 setting; falls back to right ⌘ | Built in – nothing to download | Yes, in the setup window |
| Wispr Flow | Yes – browser sign-in first | Accessibility → Microphone, cards advance on their own | Only for Caps Lock | fn (Ctrl+Opt if no fn key) | Says fn alone blocks the system action | Cloud – none | Yes: mic test and demos |
| Willow Voice | Yes – create an account | Microphone → Accessibility | Not mentioned | fn | Asks you to set “Press 🌐 key to → Do Nothing” | Cloud – none | Not documented |
| Aqua Voice | Yes – sign in | Not documented | Not documented | fn | Says the key “will not bubble down” to apps | Cloud – none | Not documented |
| VoiceInk | No (7-day trial at the end) | Microphone → Accessibility → Screen Recording (optional) | No | None – you record one | No guidance | Parakeet, 494 MB – blocks Continue until done | Yes, in onboarding |
| Superwhisper | No for local use | Microphone → Accessibility | Not mentioned | ⌥Space | Avoids fn; warns fn may go to the system | Cloud by default; local on first use | No – first real dictation |
| Spokenly | No for local use | Microphone first (one per screen) | Not documented | Right ⌘ | Avoids fn | Pick and download a local model | Not documented |
| MacWhisper | Licence – dictation is Pro | Guided from the Dictation button | Not documented | You set one | Not documented | Not documented | Not documented |
Sources: docs.wisprflow.ai (setup guide, permissions, hotkeys), help.willowvoice.com (setup, hotkey settings), aquavoice.com (guide, FAQ), VoiceInk source on GitHub (Features/Onboarding, ShortcutMonitor.swift), superwhisper.com/docs (quickstart, settings, shortcuts), spokenly.app (docs) and its first-run screen, docs.macwhisper.com (dictation). Checked 8 October 2026.
What we learned
- Microphone and Accessibility are enough for listening, pasting and the hotkey. Input Monitoring adds a third prompt and often a restart, and the apps with the smoothest setup skip it.
- The fn key is an easy trap to miss. Wispr Flow, Aqua Voice and Willow default to fn, but macOS already uses 🌐/fn for input language, emoji or its own dictation. Only Willow tells you to change the system setting. Spokenly and Superwhisper avoid the problem with right ⌘ and ⌥Space.
- The cloud apps (Wispr Flow, Aqua Voice and Willow) want you signed in before anything else. With the local apps you can dictate without an account.
- The model download takes longest. VoiceInk won't let you continue until its 494 MB model is in place, while cloud apps skip this step. FnFlow ships a model inside the installer and fetches the larger one (874 MB) in the background.
- Wispr Flow and VoiceInk let you dictate into a field during onboarding, so you see it working before going back to your own apps.
How FnFlow's setup works
- A single “Grant Access” button asks for the microphone and then Accessibility, and the window notices each grant by itself. FnFlow doesn't need Input Monitoring or a restart.
- FnFlow reads your 🌐 key setting. If fn is free it uses fn, and if not it uses right ⌘, so you don't have to open Keyboard settings. You can switch keys in Settings at any time, and shortcuts like ⌘C pressed with right ⌘ never start a dictation.
- A Whisper model ships inside the installer, so the first dictation works offline. The more accurate model downloads in the background on Wi-Fi.
- Before you leave setup, you can hold the key, say a few words and see them appear in the setup window.
- There's no account to create.
Questions
Which permissions does a Mac dictation app really need?
Two: Microphone to hear you, and Accessibility to paste the text and notice the hotkey. Input Monitoring isn't needed when the app uses an active event tap. VoiceInk and FnFlow work without it (Willow's help doesn't mention it), and Wispr Flow needs it only for a Caps Lock hotkey. Screen Recording is optional and only matters for apps that read the screen for context.
Why doesn't holding fn work for dictation?
Because macOS gives the 🌐/fn key its own job: switching input language, showing emoji or starting Apple's dictation. Either set Keyboard → “Press 🌐 key to” → “Do Nothing”, or use an app that picks another key, such as right ⌘ or ⌥Space.
Do I need to restart the app after granting permissions?
Usually not. Accessibility takes effect immediately, and apps that poll for the grant (FnFlow, VoiceInk, Wispr Flow) carry on by themselves. Input Monitoring is the one that often needs a restart, which is another reason to avoid it.
Which apps need an account before the first dictation?
The cloud-based ones. Wispr Flow, Aqua Voice and Willow ask you to sign in first, while the local apps (FnFlow, VoiceInk, Superwhisper, Spokenly) work without an account.
How did you collect this?
From each vendor's documentation and help center, VoiceInk's open-source code (onboarding and shortcut monitor) and our own look at Spokenly's first screen, on 8 October 2026. “Not documented” means we found no primary source. We make FnFlow, so read our row with that in mind.