Raspberry Pi 4 Model B // On-Device Voice Assistant
Rush Butler — Control & Voice Console
100% on-device wake word, Vosk speech-to-text, Piper neural voice synthesis, Spotify Connect, WiFi presence greeter, and executive workflow automation.
Wake Word: "Hey Jarvis" / "Rush"
Rush Butler Ready (Idle)
Latency: 180msRush Butler Chat Thread
Piper Neural TTS
Hello, I'm Rush. How may I assist your inquiry today?
Sample Voice Commands (Click to Test):
systemd journalctl -u pi-butler -u pi-greeter -f
[11:38:02]INFO[pi-butler]Vosk 16kHz speech recognition model loaded (models/vosk-model-small-en-us-0.15)
[11:38:04]INFO[pi-butler]Piper TTS voice initialized (en_US-lessac-medium.onnx)
[11:38:08]INFO[raspotify]Spotify Connect daemon ready as "Pi Butler" (ALSA hw:1,0)
[11:38:12]INFO[pi-greeter]ARP scanner daemon bound to wlan0. Interval: 20s. Cooldown: 30m
[11:38:20]TRIGGER[pi-butler]Wake word "Hey Jarvis" primed with threshold 0.50
100% On-Device Privacy Architecture
Zero-Cloud Speech Pipeline
Unlike cloud-dependent voice hubs, Rush Butler processes wake-words, speech-to-text, and speech synthesis entirely within the local Raspberry Pi 4 CPU cores:
- >Wake Word: Configurable OpenWakeWord / Precise ("Hey Jarvis" threshold 0.5)
- >Speech-to-Text: Vosk small en-US offline model (40MB RAM footprint, ~180ms latency)
- >Voice Synthesis: Piper ONNX neural model (Lessac Medium, <100ms generation)
- >Network Greeter: Subnet ARP broadcast listener for instant presence detection
Executive AI Link & Task Sync
n8n + GeminiTelegram bot bridge powered by Google Gemini AI and n8n workflows. Automatically summarizes incoming URLs, transcripts, and voice memos into structured action digests.
Simulated Telegram Ingestion:
> URL: https://arxiv.org/abs/2405.04434
> Gemini Action: Extracted 3 executive takeaways, scheduled review reminder for 07:00 PM.
> Gemini Action: Extracted 3 executive takeaways, scheduled review reminder for 07:00 PM.
Deployment Blueprint
Hardware PlatformRaspberry Pi 4 Model B
Operating SystemRaspberry Pi OS 64-bit
Audio In/OutALSA USB Mic + 3.5mm Speaker
Media PlaybackSpotify Connect (Raspotify)