VIKO
▶ Personal AI Voice Assistant
VIKO
Your voice. Your AI. Your rules.
A personal AI assistant that listens, thinks, and speaks back — powered by Gemini Live API. Lives on your machine. Knows your voice.
VIEW ON GITHUB → EXPLORE FEATURES ↓
Why VIKO
Built for one reason:
a truly personal AI.

Most AI tools talk at you through a chat box. VIKO was built to talk with you — in real-time, by voice, naturally.

The idea was simple: what if your AI assistant actually knew who you were? Recognized your voice. Remembered your conversations. Lived on your machine — not on a server farm somewhere.

VIKO is the result of that idea. A futuristic HUD on your screen. A voice that listens. An AI that can even update itself when you ask it to.

20+
Built-in skills — web search, reminders, dev agent, and more
100%
Voice-first — every feature accessible by speaking
Live
Real-time streaming audio via Gemini Live API
Capabilities
What VIKO can do
From everyday tasks to full project development — all accessible by voice, all running locally.
◉
REAL-TIME VOICE
Gemini 3.1 Live API with Silero VAD. Configurable female Indonesian voice — Erinome by default. Low-latency, no chunking.
◈
FUTURISTIC HUD
Sci-fi dark UI with live system metrics, animated world map, clock, and timestamped activity log with tool status indicators.
◑
SPEAKER VERIFICATION
Enrolls your voice using resemblyzer embeddings. Only the owner's voice unlocks VIKO — everyone else is gated.
◎
SELF-MODIFICATION
Tell VIKO to add a new skill, fix a bug, or update its own personality — and it does it, live.
◇
LONG-TERM MEMORY
Remembers your conversations and preferences. Vector memory via ChromaDB + SQLite + Gemini embeddings.
◆
20+ SKILLS
Web search, file management, app control, weather, flights, reminders, news, and more — all by voice.
◌
DEV AGENT
Describe a project out loud. VIKO plans, writes, tests, and commits the code — plan → code → git.
◐
LIVE LOCATION
GPS via macOS CoreLocation with Nominatim geocoding. Animated map marker on the HUD at all times.
Preview
See it in action
Three modes, one interface — built for a future that feels like now.
Dashboard
DASHBOARD Live metrics, GPS map, system status — always visible on the HUD.
Text Chat
TEXT CHAT Full conversation history with activity log and context awareness.
Voice Chat
VOICE CHAT Real-time audio visualizer while VIKO listens and speaks back.
Open Source
Explore the full project
on GitHub →
Source code, setup guide, architecture docs, and contribution notes — everything you need to understand or run VIKO yourself.
Gemini Live API Claude Sonnet PyQt6 Python 3.11+ resemblyzer ChromaDB SQLite macOS
VIEW ON GITHUB →