# Brainstorming a system
**URL:** https://community.openconversational.ai/t/brainstorming-a-system/18551
**Category:** SmartBoomBoxes
**Created:** [May 30, 2025, 2:21pm UTC](https://community.openconversational.ai/t/brainstorming-a-system/18551 "2025-05-30T14:21:44Z")
**Posts on this page:** 4
**Page:** 1
### Author:  [@mike99mac](https://community.openconversational.ai/u/mike99mac)
#### Post date: [May 30, 2025, 2:21pm UTC](https://community.openconversational.ai/t/brainstorming-a-system/18551/1 "2025-05-30T14:21:44Z")
Brainstorming a design for music boxes in perhaps a living room, an office, a kitchen and 3 bedrooms:
All boxes have:
- Superior craftsmanship and fidelity
- Quality microphones
- A personal voice assistant on an SoC
The box in the living room (“the hub”):
- Has a powerful SoC (e.g. Nvidia Jetson Nano)
- Is on WiFi and connected to the Internet
- Runs a STT server locally (e.g. whisper)
- Runs an AI-bot locally for “fallback” questions (best one to start with?)
- Can get AI-bot answers from the Internet
- Has music library in local file system
- Can get music and news from the Internet
The other 5 boxes:
- Have less powerful SoCs (e.g. RasPi 5)
- Are on WiFi but can only connect to the hub (iptables)
- Do STT from the hub (“remote”)
- Do TTS locally (e.g. piper)
- Get “fallback” answers from the Hub
- Mounts the music file system from the hub
- Can stream music/news w/hub as a proxy
Privacy/security advantages:
- Only the hub faces the Internet
- No speech leaves the building
- AI-bots are done locally unless asked to use Internet
- Music/news can be done locally and remotely
Has anyone got an AI-bot running locally? Any suggestions?
---
### Author:  [@goldyfruit](https://community.openconversational.ai/u/goldyfruit)
#### Post date: [May 30, 2025, 7:50pm UTC](https://community.openconversational.ai/t/brainstorming-a-system/18551/2 "2025-05-30T19:50:01Z")
Hey @mike99mac
Looks like a HiveMind scenario to me.
I did a video about running everything locally on Jetson Orin Nano a bit more than a month ago so its definitely possible but not super easy at first mostly because JetPack 6.2 sucks and I added a layer of complexity with Kubernetes. 😛
> [@rocket Open Voice OS on Jetson Orin Nano: Offline AI Assistant with LLM + TTS + STT on K3s!](https://community.openconversational.ai/t/open-voice-os-on-jetson-orin-nano-offline-ai-assistant-with-llm-tts-stt-on-k3s/18194):
>
> Hello the community! In this video, we demonstrate Open Voice OS running a fully offline AI voice assistant stack on the ultra-compact NVIDIA Jetson Orin Nano, orchestrated with K3s and managed by Rancher. Get a glimpse of a truly modern, edge-native AI assistant powered entirely on-device, featuring: brain Local LLM (Large Language Model) for smart, private conversations speaking_head High-quality Text-to-Speech (TTS) with natural voice generation studio_microphone Fast and accurate …
Since then I was able to run a newer version of Ollama which now allows me to run Gemma3, Qwen3, DeepSeek, etc… LLMs.
---
### Author:  [@mike99mac](https://community.openconversational.ai/u/mike99mac)
#### Post date: [May 30, 2025, 8:13pm UTC](https://community.openconversational.ai/t/brainstorming-a-system/18551/3 "2025-05-30T20:13:50Z")
Nice work! You’re way ahead of me. I am enjoying learning. Yes, I agree with you on JetPack. The humor skill could use a little work - maybe we can find a stand-up comedian willing to do FOSS. 😆
---
### Author:  [@goldyfruit](https://community.openconversational.ai/u/goldyfruit)
#### Post date: [May 30, 2025, 8:50pm UTC](https://community.openconversational.ai/t/brainstorming-a-system/18551/4 "2025-05-30T20:50:06Z")
> [@mike99mac](#):
>
> The humor skill could use a little work
It is not a skill but the local AI making a joke 😃