Local Voice
Real-time offline voice assistant for Pi and Linux.
By m15-ai
- Est. cost
- Cost unknown
- Posted
- GitHub stars
- 23 (GitHub)
Overview
A lightweight, low-latency voice assistant that runs fully offline on a Raspberry Pi 5 or Linux machine, with no cloud and no API keys.
How it works
PyAudio captures the microphone, Vosk transcribes in real time, and a small local LLM through Ollama (for example gemma2:2b or qwen2.5:0.5b) replies. Piper TTS speaks the answer, with optional SoX effects. A JSON config sets the model, voice and how many past exchanges the model remembers.
Hardware
- Raspberry Pi
This page is a short summary written by HardwareLab. Build steps, code and photos belong to the creator; follow “View original” for the full project.
More in Voice Assistants & Companions
See all 32esp_ai (fully offline ESP32-S3)
Speech-to-text, a tiny LLM and TTS all on one ESP32-S3.
- Freenove FNK0104B ESP32-S3
AdvancedCost unknown40
Daedalus voice terminal
An ESP32 voice terminal for a home AI running OpenClaw.
- ESP32
- mic
- speaker
- +2 more
IntermediateCost unknownRIO XIAO ESP32 S3 Gemini Live Voice Assistant
Standalone full-duplex Gemini Live assistant on a XIAO.
- XIAO ESP32-S3
IntermediateCost unknown2