WHISPER · PIPER · JARVIS

Local Whisper + Piper — Voice for Vibe Coding

VibeFuse ships whisper-cli.exe and piper.exe under %LOCALAPPDATA%\VibeFuse. Hold the title-bar mic, or say “Jarvis.” Transcription and speech stay on your PC by default. Remote hosts are fallback only.

Download with voice engines Local Servers docs

Controls that matter

Whisper models

Tiny → Small English, quantized or full. Changing the model restarts the local server on port 8935.

Piper voices

Jarvis, Lewis, Michael, Adam, Heart, Bella, Emma. Speed 0.5–2.0×, volume to 150%.

Wake word

Settings → Voice. “Yes, sir?” ack, barge-in, audio ducking, 12s command timeout.

Notes mode

Click-to-record, summarize, upload to Account → Notes. Wake word pauses.

Endpoints

Piper 127.0.0.1:4123 · Whisper 127.0.0.1:8935 — editable if you run your own.

Sister app

Need system-wide dictation? VocalFuse is the paid overlay. Same account. Comparing built-ins and paid dictation apps? See the best free dictation software for Windows comparison.

✓ On-device Whisper
✓ On-device Piper
✓ Jarvis wake word
✓ Notes upload
✓ No voice subscription

Install local voice Free STT guide

Explore VibeFuse & harness guides

Local AI voice FAQ

Does VibeFuse use Whisper and Piper locally?

Yes. whisper-cli.exe and piper.exe live under %LOCALAPPDATA%\VibeFuse\bin with models in ...\models. Remote hosts are fallback only.

What is the Jarvis wake word?

Enable it in Settings → Voice. Say “Jarvis”; VibeFuse plays “Yes, sir?” and records your command. Local-first STT/TTS.

Can I change the Whisper model?

Yes. Settings → Local Servers → Whisper model. Tiny is fastest; Small is more accurate. Changing it restarts the local server.

Is this the same as VocalFuse?

Related engines, different product. VibeFuse voice drives the coding canvas. VocalFuse is the paid system-wide dictation overlay.