Initial commit: HA Voice Control MCP server
This commit is contained in:
@@ -0,0 +1,75 @@
|
||||
# Deployen op Synology NAS
|
||||
|
||||
## Stap 1: Code naar je NAS kopiëren
|
||||
|
||||
```bash
|
||||
# Vanaf je Windows machine:
|
||||
scp -r C:\Users\moel-\Desktop\dev\mcp mo@192.168.1.211:/volume1/docker/ha-voice-control/
|
||||
# Of via SMB: \\192.168.1.211\docker\ha-voice-control\
|
||||
```
|
||||
|
||||
Je hebt deze bestanden nodig op de NAS:
|
||||
- `Dockerfile`, `docker-compose.yml`, `.dockerignore`
|
||||
- `config.py`, `requirements.txt`, `requirements-neo4j.txt`
|
||||
- `src/` (volledige map)
|
||||
- `static/` (volledige map)
|
||||
|
||||
## Stap 2: .env aanmaken op de NAS
|
||||
|
||||
```bash
|
||||
ssh mo@192.168.1.211
|
||||
cd /volume1/docker/ha-voice-control
|
||||
nano .env
|
||||
```
|
||||
|
||||
Inhoud van `.env`:
|
||||
```env
|
||||
HA_TOKEN=jouw-home-assistant-token
|
||||
PG_PASSWORD=WaQTUw2t
|
||||
NEO4J_PASSWORD=WaQTUw2t
|
||||
```
|
||||
|
||||
## Stap 3: Bouwen en starten
|
||||
|
||||
```bash
|
||||
cd /volume1/docker/ha-voice-control
|
||||
|
||||
# Bouw de image (duurt ~5 min eerste keer)
|
||||
docker-compose build
|
||||
|
||||
# Start de container
|
||||
docker-compose up -d
|
||||
|
||||
# Check logs
|
||||
docker-compose logs -f
|
||||
```
|
||||
|
||||
## Stap 4: Testen
|
||||
|
||||
```bash
|
||||
curl http://localhost:8765/api/health
|
||||
curl http://localhost:8765/api/dashboard/overview
|
||||
```
|
||||
|
||||
Dashboard: `http://192.168.1.211:8765/dashboard`
|
||||
Voice Control: `http://192.168.1.211:8765`
|
||||
|
||||
## Optioneel: achter nginx/reverse proxy
|
||||
|
||||
Als je de interface via je bestaande domein (`ha.el-kadi.nl`) wilt:
|
||||
- Voeg een proxy_pass toe in je nginx config naar `http://localhost:8765`
|
||||
|
||||
## Whisper model aanpassen
|
||||
|
||||
In `docker-compose.yml`, pas `WHISPER_MODEL` aan:
|
||||
- `tiny` (~75MB) — snelste, voor simpele commando's
|
||||
- `base` (~145MB) — aanbevolen voor Nederlands
|
||||
- `small` (~488MB) — betere herkenning, trager op NAS CPU
|
||||
|
||||
## Container Manager (DSM UI)
|
||||
|
||||
Als je DSM's Container Manager gebruikt i.p.v. CLI:
|
||||
1. Open Container Manager → Project → Create
|
||||
2. Project name: `ha-voice-control`
|
||||
3. Selecteer `/volume1/docker/ha-voice-control/docker-compose.yml`
|
||||
4. Klik Next → Done
|
||||
Reference in New Issue
Block a user