Smarter writing, in one shot.
Just talk.
Hold a key, talk, let go. The cleaned-up text lands at your cursor. No copy, no “rewrite this” prompt, no paste. The model is bundled. Nothing leaves your Mac.
First 100 get $39 at launch.Pay once, use forever.
Your words never leave your Mac.
Other apps stream your voice to a server they own. We run the whole pipeline on your Mac, so nobody has to take our word for it.
Other dictation apps
- You talk into your mic
- Audio uploaded to their server
- Their model runs in their data center
- Their team can review it. Stored on their disk.
Transcribr Flow
- You talk into your mic
- Local Whisper model runs on your Mac
- Cleanup AI runs on your Mac too
- Audio + text sit on your disk. No upload, ever.
No cloud calls. No analytics. No telemetry. Pull the plug on your Wi-Fi and it still works. Airplane mode is fine. That's not a marketing line. It's how we built it.
See it in motion
Voice in. Clean text out.
A live look at the four phases. Hold the shortcut, talk, watch words land, then watch the cleanup tidy them up. The whole loop runs on your Mac.
You talk.
Hold the shortcut, talk into your Mac. Audio never leaves the laptop.
Works everywhere
One shortcut. Every app you type into.
Transcribr Flow is not another writing app you switch into. Push the shortcut from wherever your cursor already is. The cleaned text lands right there. Mail, your editor, the URL bar, a Slack reply, a Notion page. If you can type in it, you can talk into it.
What it looks like in practice
- Pasted into MailYou said
“send the proposal to legal by friday”
Typed for youSend the proposal to Legal by Friday.
- Pasted into SlackYou said
“merged the auth fix can you review”
Typed for youMerged the auth fix. Can you review?
- Pasted into NotionYou said
“spec update we're moving the onboarding step to after signup”
Typed for youSpec update: we're moving the onboarding step to after signup.
- Pasted into CalendarYou said
“lunch with mom tuesday at noon”
Typed for youLunch with mom, Tuesday at noon.
App names and logos are trademarks of their respective owners. Transcribr Flow is not affiliated with or endorsed by any of them. It works with them because macOS lets text input come from anywhere.
The cleanup philosophy
Your voice. Just better written.
Most AI tools rewrite your message in their voice. Transcribr Flow fixes what’s broken (grammar, capitals, punctuation, flow) and leaves what isn’t. You wrote it. We just polished it.
hey could you send me the report once you're done i also need the spreadsheet from priya by friday thanks
Hey, could you send me the report once you're done? I also need the spreadsheet from Priya by Friday. Thanks!
Same words. Same intent. Same voice. Punctuation added, capitalization fixed, no rewriting.
More things it cleans
- GrammarYou said
the report and the spreadsheet is ready
Transcribr Flow typedThe report and the spreadsheet are ready.
- CodeYou said
chmod six four four dot env
Transcribr Flow typedchmod 644 .env
- Names and timeYou said
schedule a call with priya tuesday at three
Transcribr Flow typedSchedule a call with Priya Tuesday at 3.
Zero setup
The cleanup model ships inside the app.
v1.0 bundles a small local LLM (qwen3 1.7B, running via Apple’s MLX framework). No Ollama to install, no account to create, no key to paste. Open the app and it works. If you want to bring a larger local model later, the Direct version lets you point it at your own Ollama setup from Settings.
No account. No subscription. No data leaves your Mac.
Who it’s for
Built for the way you actually work.
Four people, four reasons to keep dictation local. Tap any card for the longer version.
Multilingual · 20+ languages
Talk in your language. Get text in your language.
Whisper recognizes a hundred languages well. Transcribr Flow exposes the ones people actually dictate in. Transcription runs locally in every language. The bundled cleanup LLM covers Latin-script ones out of the box.
Plus ~80 more in the model picker. Switch the recognition language per recording in Settings, or let auto-detect handle it.
Under the hood
Open where it should be. Closed where it matters.
We don't invent privacy from scratch. We assemble it from tools you can audit yourself, then add the layer that makes them feel like one app on your Mac.
Public projects with permissive licenses. Read the source, run the models yourself, verify the privacy claim end to end.
- WhisperMIT
The open-source speech model behind WhisperKit. Runs locally on your Mac. The audio never leaves the machine.
- MLXMIT
Apple's machine learning framework for Apple Silicon. Powers the bundled cleanup LLM and runs it on the Neural Engine.
- Apple SiliconPlatform
Every model path is tuned for the Neural Engine and Metal. M-series Macs only, by design.
The layer that turns four open projects into one product that feels native. Our work, our code, our continued investment.
Developer-vocabulary tuning
A terminology layer for code, CLIs, library and identifier names so paths and camelCase survive transcription without manual fixes.
Push-to-talk pipeline
Sub-second key-down to paste across any focused app. Capture, transcription, cleanup, and insertion stitched into one perceived action.
Cleanup orchestration
Context-aware prompting that fixes grammar, punctuation, and capitalization without rewriting your voice. The model is yours; the conductor is ours.
macOS integration
Accessibility paste, focus tracking, system shortcuts, and a recordings library built directly against Apple frameworks.
Project names and logos are trademarks of their respective owners. Transcribr Flow is built with these projects but is not affiliated with or endorsed by them.
How we differ
Wispr is great. Just not on your machine.
We run the model where the audio is. They run it where their customers are. The trade-off is yours. Here are the facts.
OursTranscribr Flow | Wispr Flow | MacWhisper | |
|---|---|---|---|
| Transcription runs | On your Mac | Their servers | On your Mac |
| Cleanup runs | Bundled local LLM (no setup) | Their servers | Local Ollama or cloud |
| Subscription required | No | Yes for full features | No |
| URL transcription (YouTube, IG, TikTok) | Yes (direct release) | No | YouTube only |
Pricing and features change. Always verify on each vendor's site.
Pricing
Buy it once. Keep it forever.
One price on both channels: $59 lifetime. No subscription, no renewal, no Pro tier in v1.0. Pick the SKU that fits how you want to buy.
Requires Mac with Apple Silicon (M1 or later) · macOS 14 or newer
On an Intel Mac or Windows? Email support@devxgroup.io. We track demand by hand and we'll let you know if we ship it.
ONE-TIME
App Store
Apple bills you. Simple, sandboxed, and local.
$59lifetime- Bundled local LLM cleanup (no installs)
- Local Whisper transcription (bundled model)
- Language picker
- Transcript export (TXT, SRT, VTT)
- In-window hold-to-talk shortcut
- Menu bar item
- Start Dictation App Intent for Shortcuts.app
The sandboxed version. Apple bills you, App Store handles updates. Good if you want the App Store-verified experience.
Buy on App Store- Full power
FOUNDER PRICING · FIRST 100
Direct
Full-power version. Global shortcuts, URL ingestion, auto-paste.
$39lifetimeSave $20 · First 100 customersthen $59
- Everything in the App Store version
- Global hold-to-talk from any app
- Auto-paste into focused field after dictation
- Streaming dictation (types as the LLM produces)
- YouTube, Instagram, TikTok URL ingestion
- ffmpeg conversion
- Advanced: bring your own local Ollama model
- Updates install themselves (one click)
- Use on up to 3 of your own Macs
- All sales final (statutory rights preserved)
The full-power version. Uses Input Monitoring and Accessibility to reach any app on your Mac.
Compare every feature
What you get at each SKU.
Both SKUs bundle a local LLM and run everything on your Mac. The direct version adds features that need permissions Apple's sandbox doesn't allow. Same price either way.
| Feature | App Store$59 lifetime | DirectFull power$39 lifetime |
|---|---|---|
| Both SKUs. | ||
| Bundled local LLM cleanup (no installs) | ||
| Local Whisper transcription (bundled model) | ||
| Language picker | ||
| Transcript export (TXT, SRT, VTT) | ||
| In-window hold-to-talk shortcut | ||
| Menu bar item | ||
| Start Dictation App Intent (Shortcuts.app) | ||
| Window menu reopen | ||
| Direct only. | ||
| Global hold-to-talk from any app | ||
| Auto-paste into focused field after dictation | ||
| Streaming dictation (types as LLM produces) | ||
| YouTube, Instagram, TikTok URL ingestion | ||
| ffmpeg conversion | ||
| Advanced: use your own local Ollama model | ||
| Updates install themselves | Via App Store | In-app, one click |
| License works on | 1 Mac | 3 Macs |
| Refund window | Apple's policy | All sales final |
Why these prices
We’re honest about what costs us money and what doesn’t.
- 01
Why $59 on both
Same price, both channels. One purchase, lifetime updates within version 1. No subscription, no hidden tier.
- 02
Why $39 founder
First 100 customers get $39 as a thank-you for buying before reviews and word-of-mouth catch up. After that, it's $59.
- 03
Why two SKUs
The App Store version runs in Apple's sandbox, which is stricter by design. The direct version uses Input Monitoring and Accessibility to reach any app on your Mac. Same price, different permissions model.