Up and running in two minutes.
braindumpi is a native Mac menu-bar app. Install it, grant a few permissions, and you're dictating into compatible text fields. Here's the whole flow.
braindumpi is a native Windows app that lives in your notification area. Install it, allow the microphone, and you're dictating into compatible text fields. Here's the whole flow.
01Install braindumpi
- Download Braindumpi.dmg from braindumpi.com/download.
- Open the downloaded .dmg, then drag the braindumpi icon into your Applications folder.
- Eject the disk image and launch braindumpi from Applications (or Spotlight).
braindumpi runs on Apple silicon with macOS 14 and later. Closing its window leaves it in the menu bar, ready for your next dictation. Use Quit to stop the app.
- Download Braindumpi-Setup.exe from braindumpi.com/download.
- Run it. Windows will show a blue "Windows protected your PC" box because the installer is not code-signed yet — click More info, then Run anyway.
- The installer sets braindumpi up for your account only, so it never asks for an administrator. It launches when it finishes.
braindumpi runs on 64-bit Windows 10 version 1809 and later. It lives in your notification area — there's no taskbar window to manage.
02First launch & sign-in
On first launch, braindumpi opens the Hub. Sign in with your email to sync your plan across every machine you use — the same account works on Mac and on Windows, so if you already paid on one, just sign in with the same email on the other. You can dictate on the free tier right away — no card required. Your one-time welcome week includes 7 days of Pro, followed by Free: 1,000 words per week, slower dictation before a 2,000-word cap, and a one-time 1,000-word gift. Dictionary, snippets, and settings stay on each device.
03Grant permissions
macOS asks for three permissions so braindumpi can hear you and type at your cursor. Approve each when prompted, or set them under System Settings → Privacy & Security.
If dictation isn't landing, it's almost always a permission that got skipped. Re-check Accessibility and Input Monitoring in System Settings, then quit and reopen braindumpi.
Windows asks for one permission: the microphone. Allow it when prompted, or turn it on under Settings → Privacy & security → Microphone. There is no equivalent of the Mac's Accessibility or Input Monitoring consent, so there is nothing else to approve.
Make sure "Let desktop apps access your microphone" is on as well — it sits further down the same Settings page, and it is off on some machines even when the main switch is on.
04Push to talk
Put your cursor wherever you want text — an email, a chat box, your editor, a ChatGPT prompt. Then:
- Hold the fn / 🌐 globe key.
- Hold the right ctrl key.
- Talk normally. Ramble, correct yourself, whatever — the pill overlay shows it's listening.
- Release the key. When transcription and cleanup finish, the result appears in the active editable field. Processing time depends on the recording and your computer.
The push-to-talk key defaults to the Fn / globe (🌐) key. You can change it in the Hub's Voice settings if you'd rather use a different key.
The push-to-talk key defaults to right Ctrl. Change it under Settings → Shortcuts → Dictation. Right Alt can be reserved for typing on keyboards that use AltGr.
05The pill menu
While braindumpi runs, a small pill / orb overlay sits at the bottom of your screen. Hover it to open the quick menu, where you can:
- Paste last dictation — drop your most recent result again.
- Language control — switch between Auto (detect the language you speak) and Fixed (lock to one language).
- Open the Hub for your dictionary, snippets, and settings.
06Paste the last dictation
Need your last result somewhere else? Press Ctrl Alt V to paste the most recent dictation at your cursor — no need to re-record. It's the same action as "Paste last dictation" in the pill menu.
Need your last result somewhere else? Press ⌥ ⌘ V to paste the most recent dictation at your cursor — no need to re-record. It's the same action as "Paste last dictation" in the pill menu.
07Switching languages
braindumpi handles multiple languages. In Auto mode it detects what you speak and writes it back in the same language. Set Fixed to lock one language when auto-detect keeps guessing wrong. Both modes live in the pill's language control and keep dictation in the original language.
Choose Automatic or a fixed input language under Settings → Speech language. Dictation stays in the original language.
Automatic translation is temporarily unavailable because the local model did not meet our accuracy checks. Dictation in the original language and explicit Command remain available.
Local writing AI and Command
On Mac and Windows, the optional Qwen3 4B writing model is a separate download of about 2.50 GB. It processes text on your computer. Command and AI cleanup require Pro, including during the welcome week. Basic rule-based cleanup works without this download.
Download and enable the model in Settings → Local cleanup & Command Mode. To edit, select text in an editable field, hold the Command shortcut shown in Settings, say the change, then release.
Download Local writing AI — Qwen3 4B (2.50 GB, optional) in Models. To edit, select text in an editable field, hold the shortcut under Settings → Shortcuts → Edit selected text (Pro), say the change, then release. Set Settings → Default cleanup → High (local AI, Pro) to use it for dictation cleanup.
Dictation cleanup only changes punctuation, capitalization, and clear vocal hesitations. It keeps your words in order, including repetitions and self-corrections. Tone does not change dictation style automatically. Use Command for an explicit rewrite. If the model is unavailable or fails the content checks, cleanup falls back to local rules. Review Command results before using them. Team sharing is not available.
08Cloud and offline
braindumpi transcribes through the cloud by default. Audio streams straight from your computer to our speech provider during recording. The cloud handles speech recognition; cleanup and the optional writing AI run on your computer.
A separate speech model downloads for offline transcription. Offline dictation is available after that download completes. The optional Qwen3 writing model is a different download and does not transcribe audio.
To keep transcription audio on your computer, finish downloading the speech model and turn Cloud transcription off in the Hub. Account sign-in and sync, updates, and model downloads still use the network. Both platforms also offer an optional typing reminder in Settings; it counts key presses and timing in memory without capturing typed characters.
09Upgrade to Pro
The Free allowance is 1,000 words a week. For unlimited dictation, upgrade from the Upgrade button in the Hub:
- $6/month, $32/year (best value), or $470 one-time for lifetime.
- Checkout is handled by Polar. Monthly has a 7-day trial and annual a 14-day trial for eligible customers, both with a card and automatic renewal. Lifetime charges once at checkout. Cancel in Billing before the trial or paid period ends to prevent the next renewal. Every paid plan has a 14-day refund window.
See full plan details on the pricing section.
10Get help
Stuck on something not covered here? Email support@braindumpi.com. For how we handle your data, see the Privacy Policy; for plan and refund terms, see the Terms of Service.