How to Build a Talking AI (Offline, Open-Source & Free)

What if your AI assistant didn't need the internet — or cost anything to run?

In this video, we break down exactly how to build a fully offline,
open-source talking AI from scratch — the same pipeline powering
real robots like the PiDog on a Raspberry Pi.

✅ What you'll learn:
00:00 - How a talking AI actually works (the full pipeline)
01:30 - Meet the PiDog: offline voice chatbot on a Raspberry Pi
02:30 - The 3 building blocks: Ears → Brain → Mouth
03:00 - Vosk: offline speech-to-text (the ears)
04:00 - Ollama: running an LLM locally (the brain)
05:00 - Piper: text-to-speech (the mouth)
06:00 - WebSockets vs WebRTC — which one for live audio?

🔧 Tools used (all free & open-source):
→ Vosk: https://alphacephei.com/vosk
→ Ollama: https://ollama.ai
→ Piper TTS: https://github.com/rhasspy/piper

🔔 Subscribe for more CS & AI concepts explained visually.
👈 Watch next: Object-Oriented Programming Explained (Animated)

#AITutorial #VoiceAI #Ollama #Vosk #RaspberryPi #OpenSourceAI #BuildAI

Видео How to Build a Talking AI (Offline, Open-Source & Free) канала Programmingwithneda