VoxPilot v0.4.6: Windows Audio Capture Fix

VoxPilot v0.4.6 — Windows Audio Capture Fix

This release fixes a critical bug that prevented VoxPilot from capturing audio on Windows.

What changed

The previous Windows audio path used audio=default as the ffmpeg dshow input — but dshow does not support a generic "default" device name. It needs an actual DirectShow device name like "Microphone (Realtek Audio)".

Fixes in v0.4.6:

• Auto-detect microphone — when no device is configured, VoxPilot now queries ffmpeg for available DirectShow audio devices and picks the first one automatically
• PowerShell fallback — if ffmpeg device listing fails, falls back to Get-CimInstance Win32_SoundDevice to find input devices
• Fixed binary detection — which doesn't exist on Windows; now uses where so ffmpeg is actually found in PATH
• Cleaner output — added -hide_banner -loglevel error flags to suppress noisy ffmpeg startup text

By the numbers

🎙️ 4,346 downloads on Open VSX and growing.

Get it

• Landing page: natearcher-ai.github.io/voxpilot
• Open VSX: open-vsx.org/extension/natearcher-ai/voxpilot
• GitHub: github.com/natearcher-ai/voxpilot

Update in Kiro or VS Code to get it. If you're on Windows and voice input wasn't working before — give it another shot.

---

VoxPilot is a free, open-source, fully on-device voice-to-code extension. No API keys, no cloud, just your voice.