WhisperLiveKit
WhisperLiveKit is available in the NetActuate Marketplace as a one-click, open-source appliance you can deploy into any NetActuate PoP. Real-time low-latency speech-to-text: a live demo page plus a WebSocket API. It is grouped under the Media & Streaming category.
What It Does
WhisperLiveKit is an open-source, real-time speech-to-text engine for low-latency live transcription. It streams audio in over a WebSocket and returns incremental transcripts using Whisper-family models with a simultaneous-streaming (LocalAgreement / SimulStreaming) policy, so words appear as they are spoken rather than after the utterance ends.
Key features
- Real-time, incremental transcription over a simple WebSocket (/asr)
- Bundled browser demo page for instant live transcription in your browser
- Whisper 'tiny' model pre-baked, so the appliance serves immediately with no download
- Voice-activity detection and streaming buffering tuned for low latency
- Deepgram-compatible and OpenAI-style transcription endpoints for easy integration
- Runs fully on CPU, no GPU required
Use cases
- Live captioning for meetings, streams, and events
- Real-time transcription behind your own applications and pipelines
- Edge speech-to-text near your users to cut round-trip latency
Why run it on the NetActuate edge Real-time transcription is latency-sensitive: hosting WhisperLiveKit in a PoP close to your speakers shortens the audio round-trip and keeps voice data on infrastructure you control.
Deployed from the NetActuate marketplace, WhisperLiveKit boots pre-configured: open the demo page at / to start transcribing, or point a client at the /asr WebSocket. The 'tiny' model is pre-baked and loads offline; selecting a larger model downloads it from Hugging Face at first boot. The appliance serves plain HTTP with no authentication, so put TLS termination and access control in front before exposing it publicly.
Vendor and Project
WhisperLiveKit is an open-source project, offered here under the Open Source listing with best-effort community support.
- Project home: github.com
NetActuate supports the underlying infrastructure. Application-level support follows the listing SLA class (open-source, best effort).
Sizing
| vCPU | RAM | Disk | |
|---|---|---|---|
| Minimum | 2 | 4 GB | 30 GB |
| Recommended | 4 | 8 GB | 50 GB |
No VPU (hardware acceleration) is required for this appliance. Choose a size at deploy based on your expected load.
Configuration Fields
When you deploy, you provide these fields. They are written into the appliance at first boot.
| Field | Required | Default | Notes |
|---|---|---|---|
| HTTP port | No | 8000 | TCP port the web UI and /asr WebSocket listen on (default 8000). |
| Whisper model | No | tiny | Whisper model to load. 'tiny' is pre-baked and loads offline; any other value (base, small, medium, large-v3, or an HF repo id) is downloaded from Hugging Face at first boot and needs outbound network. |
Using the Appliance
Start transcribing in minutes
- Open the demo. Browse to http://YOUR_VM_IP:8000/ for the bundled live-transcription page, allow microphone access, and speak to see real-time text.
- Integrate. Stream audio to the WebSocket at ws://YOUR_VM_IP:8000/asr, or use the Deepgram-compatible / OpenAI-style transcription endpoints.
- Pick a model. The 'tiny' model is pre-baked and offline; set a larger model at deploy time to trade latency for accuracy (downloaded on first boot).
Exposure posture: the appliance serves plain HTTP on port 8000 with no login. Put a TLS-terminating proxy and access control in front before exposing it to the public internet.
FAQ
Is there a software license fee? No. WhisperLiveKit is open source and free to run. You pay only for the NetActuate infrastructure and bandwidth.
Which port does the appliance serve on?
The primary service is published on port 8000 by default; see the configuration fields for any adjustable ports.
Who supports it? NetActuate operates the underlying infrastructure. The application itself is community-supported open source (best-effort SLA class).
Version and Changelog
| Version | Status | Released |
|---|---|---|
| 0.2.24 | Current (default) | 2026-07-13 |
Deploy
- Deploy in the portal: Deploy WhisperLiveKit
- View the listing: portal.netactuate.com/platform/marketplace/l/whisperlivekit
- Public listing: netactuate.com/marketplace/whisperlivekit
Need Help?
Contact support@netactuate.com or open a support ticket from the portal.