Transcribe anywhere.
Turn on Airplane Mode. Keep talking. Tulpa keeps up.
Dictation built for productivity. No word limits, on or offline.
Speak in any app and the words appear where your cursor is, transcribed on your Mac in dozens of languages. Cleanup polishes the transcript on your Mac or through a cloud model you connect.
Works on Apple Silicon Macs • 14-day free trial
Works everywhere you already type†
- Slack
- Microsoft Teams
- Google Workspace
- Linear
- Notion
- Terminal
No text field selected? Tulpa copies your words to the clipboard and tells you, so nothing is lost.
Communication styles
A voice for every task.
Speak in any app with a text field. Tulpa turns your voice into ready-to-use words right where your cursor is.
What you say
Um, can we move our team sync to Thursday? Alex is still waiting on access, and those staging keys need rotating, you know.
What lands in your app
Can we move our team sync to Thursday? Alex is still waiting on access, and the staging keys need rotating.
0:07
The way you work
Dictate wherever work takes you.
From a quiet flight to a quick reply, Tulpa keeps up with the place and pace of your work.
Any app you can type in.
From a quick reply to a command, Tulpa follows the cursor.
Your words, your way
Make every word sound like you.
Add specialist terms and dictate naturally in the languages you use every day.
Make it sound like you.
Medical terminology, marketing shorthand, legal phrases, and even Pokémon names stay spelled your way.
MEDICAL
metoprololauscultationDr. Nguyen
MARKETING
MQLROASA/B testcampaign ID 24B
LEGAL
force majeureindemnification§ 12.4
POKÉMON TRAINER
PikachuRayquazaGengar
Speaks your language, natively.
On your Mac, there are 100 ways to say:
Hello
ENLet's ship it Friday.
ESVamos a lanzarlo el viernes.
DEWir veröffentlichen am Freitag.
THวันศุกร์นี้ส่งงานกัน
Private by design
Choose where AI runs.
Keep transcription on your Mac, then choose whether any optional cleanup uses a model you connect.
Two ways to handle AI.
Pick the policy that fits your team.
Local Only
Model API
Private by default.
On-screen waveforms4 of 6 styles
0 audio uploads
Transcription happens where you're sitting, not on someone's server.
New features
Thoughtful updates for the way you work.
We keep refining Tulpa around the moments that make dictation feel natural. Here’s a closer look at focused dictation.
In the spotlight
Dictate right into the field you’re using.
Tulpa types into your selected text field. If nothing is selected, your transcript is copied to the clipboard, ready to paste.
Claude Code•••
Message
Ask Claude to…
Selected field
Private by design
What Tulpa
never does.
No audio recordings, no screenshots, no account to get started. If you want a cloud model, you bring your own key, and Tulpa tells you plainly what changes before you connect.
| Question | Tulpa | Superwhisper* | Wispr Flow* |
|---|---|---|---|
| Where your voice is transcribed | On your Mac, always | On your Mac or in the cloud, depending on the model | In the cloud |
| Audio recordings kept | None, ever | Saved with your history | On its servers when cloud storage is on |
| Reads your screen for context | Never | Super Mode uses what you're working on | Context Awareness can send a screenshot and on-screen text |
| Cloud AI | Only with your own key, after a plain warning | Built-in cloud models, or your own key on Pro | Built in |
| Model training | No. Tulpa does not train on your audio or transcripts. | No. Says customer content is not used to train, fine-tune, or improve models.* | Optional. Audio, transcripts, and edits may improve models when enabled. Data controls |
Solo
Type less. Ship more.
- No meter, ever. Dictate as much as you want, with no per-word cost.
- Works with no connection. On a plane, on a train, in a dead-signal basement.
- Your data, your choice. True local dictation, with cleanup options Tulpa never keeps or sells.
Tulpa Solo
Teams
Team-wide compliance with a single click.
- Built for teams. Join the list for a note when Teams opens.
- Everything in Solo, on every seat. On-device transcription, no meter, and no audio that ever leaves a Mac.
- Privacy built in. On-device transcription with team-wide settings.
Tulpa TeamsFAQ
Frequently asked questions.
Does my audio ever leave the device?
Never. Transcription runs on your Mac, no code path uploads audio, and no audio is saved. Dictated text leaves only if you connect a cloud model with your own API key and turn cleanup on with it, and then it goes straight from your Mac to that provider, never through us. Security has the details.
What does it cost?
Solo is $8 a month or $69 a year after a 14-day free trial, with no card needed to start. If it is not for you, we refund within 30 days. Teams is planned at $12 per seat per month, with final details coming at launch. Words are unlimited, since local inference has no per-word cost.
Which platforms does it run on?
Macs with Apple Silicon, an M1 or later, today. Intel Macs are not supported yet. Windows and Linux share the same codebase and are on the roadmap for the coming months. Join the waitlist to hear when they land.
Do you train on my dictation?
Never. We do not train on your audio or your text. If you connect a cloud model for cleanup, the text you send it is handled under that provider's terms, which is why Tulpa warns you and asks you to confirm before it connects.
How does Tulpa compare with Superwhisper and Wispr Flow?
Tulpa focuses on local-first dictation for Apple Silicon Macs, unlimited on-device transcription, and cleanup with a model you choose. Superwhisper and Wispr Flow offer different platforms and workflows. See the comparison table, and check their current Superwhisper details and Wispr Flow plans before you decide.
Can I use Tulpa for coding, writing, and everyday messages?
Yes. Dictate wherever your cursor is. Custom words help with names and terms, and communication styles shape the message for the moment.