1
Enable sharing on the strong machine
On your powerful desktop, turn on network sharing. Voicetypr exposes its local transcription engine over your LAN through a built-in HTTP API. Add an optional password to lock the endpoint down.
VoicetyprNetwork transcription
Run the big models on your powerful machine, dictate from your light one
Network transcription lets a light laptop route dictation to a powerful desktop over your own Wi-Fi. The strong machine runs the big local model, the light one just speaks — and your audio never leaves your network.
Stays on your LANNo internet requiredLocal models onlyShips in 2.0
How it works
Three steps to dictate across your devices
2
Scan and select it on the light one
On your laptop, open the Models tab and scan for LAN servers. Add the desktop, select it, and every dictation now routes to that machine over your own Wi-Fi.
3
Dictate like nothing changed
Hold your hotkey and speak. Audio travels to the strong machine, the big model transcribes it, and the text lands in your focused field — same flow, more horsepower.
What you get
Built into the Models tab, no extra setup
- Any Voicetypr instance can expose its local transcription engine over the LAN through a built-in HTTP API.
- In the Models tab you scan for, add, and select LAN servers — picking one routes all dictation to that machine.
- An optional password protects the sharing endpoint, and you can see the real active-connection count.
- Windows firewall detection flags blocked connections and walks you through the fix.
- Only local models can be shared — cloud STT sources can't be routed over the network.
Who it's for
One strong machine, many places to dictate from
Power users with a desktop and a laptop
Keep the heavy Whisper or Parakeet models on the desktop that can actually run them, and dictate from the laptop you carry around the house.
Weak or older laptops
If your travel machine can't handle the larger local models, borrow the desktop's engine over the network instead of dropping to a smaller model.
Privacy-conscious teams on a LAN
Stand up one capable machine and let everyone on the local network route to it — without sending a single second of audio to a cloud service.
Stays on your network
Your audio stays between your own devices
This is the whole point. Network transcription happens on your LAN, between your machines — audio travels from the light device to the strong one and gets transcribed there. It never touches Voicetypr's servers or any third-party cloud. Only local models can be shared this way; cloud STT sources can't be routed over the network. If you later turn on optional cloud STT or AI text formatting, those are opt-in and separate, and AI formatting sends text only, using your own provider and key.
FAQ
Network transcription questions
Does my audio leave my network?
No. Network transcription stays on your LAN, between your own devices. Audio travels from the light machine to the strong one over your Wi-Fi and is transcribed there. It never touches Voicetypr's servers or any third-party cloud.
Do I need an internet connection?
No. You only need both devices on the same local network. Once a machine is sharing and another has selected it, dictation works fully offline — no internet required.
How do I secure the sharing endpoint?
Set an optional password on the machine that's sharing. Other devices have to provide it before they can route dictation to that engine, so a stranger on the same network can't quietly use it.
Can I share a cloud transcription source over the network?
No. Only local models (Whisper, Parakeet) can be shared. Cloud STT sources can't be routed across devices — network transcription is local-engine to local-engine, by design.
I'm on Windows and connections won't go through — what now?
Voicetypr detects when the Windows firewall is blocking the sharing endpoint and shows you how to remediate it, so the light machine can reach the strong one.
Related guides
More on local, private dictation
Compare the AI-tool workflow, the broader offline case, and how Voicetypr keeps your voice on your own hardware.
ai workflows
Voice for AI agents
Dictate prompts and context into AI tools — a natural fit once your strong machine is doing the transcribing.
See the AI agents page →
offline by default
Offline dictation
The broader case for local transcription, for buyers who care most about keeping voice on their own hardware.
See the offline guide →
privacy
Zero-knowledge voice typing
How Voicetypr is built so we never see, touch, or store your transcripts.
Read about zero-knowledge →
Big-model dictation, anywhere on your network
3-day free trial. No credit card. Pay once from $39. Your voice stays on your own devices.