sam/obsidian-vault
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Family Home Lab overview: document dsh transcription (local whisper, m4a/mp3/mp4) + Ask with a file

Browse Source
This commit is contained in:
Sam Rolfe
2026-09-12 08:53:35 +10:00
parent 75769d418a
commit 3aaa615f01
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
200 projects/Tools Software WebUI/Family Home Lab.md
View File

@@ -19,7 +19,7 @@ A self-hosted **family console + tool lab** on our home network. One login at **
## What it is
- **Console portal** — per-user logins (sam/jo/harry/finn), searchable home-dash tool cards, admin area (users, passwords, live Pi-dashboard), Files upload, dark mode.
- **AI chat (dsh)** — a private per-user chat (chat/reasoning/coding + multimodal GLM), with conversation memory, web/doc summarise, image ingest & generate, and video screen analysis. Hardened: read-only filesystem, no shell.
- **AI chat (dsh)** — a private per-user chat (chat/reasoning/coding + multimodal GLM), with conversation memory, web/doc summarise, image ingest & generate, video screen analysis, **audio transcription (speech-to-text, converts m4a/mp3/mp4/AAC locally via whisper)** and a general **Ask with a file** tool (audio/video/image/txt - the AI routes by type). Hardened: read-only filesystem, no shell.
- **Voice assistant (Jervis)** — a local LLM voice agent: talk to it or type on the Voice page — it controls Home Assistant (lights/switches), shopping list, weather, timers, reminders, character voices.
- **Media tools** — GIMP (edit), KdenLive (video), Audacity (audio), LMMS (music) — all browser-based.
- **Transcriber** — turn audio **or video** into MIDI (Basic Pitch) or sheet music (MuScriptor).
@@ -29,7 +29,7 @@ A self-hosted **family console + tool lab** on our home network. One login at **
1. Go to **console.lab.audasmedia.com.au** and log in.
2. **Dashboard** — every service is a card, grouped and searchable; kids get a reduced view, admins see everything.
3. **Files** — upload from your browser; it appears in the media tools (~/media) and you can view/download/delete.
4. **Chat** — pick a model (chat / reasoning / coding / multimodal), attach images, summarise a page or file, generate an image, or analyze a screen recording.
4. **Chat** — pick a model (chat / reasoning / coding / multimodal), attach images, summarise a page or file, generate an image, analyze a video, or **transcribe an audio/video file to text** (upload to Transcribe audio - converted locally, transcript downloadable as .txt).
5. **Voice** — type a command (lights, weather, shopping, timers) and Jervis answers and speaks.
6. **Admin** — add/rename users, change passwords, watch live agent sessions.
@@ -65,7 +65,7 @@ flowchart LR
## Tools used
- **Web/portal** — FastAPI · Jinja2 · HTMX · Postgres · Redis · Celery/RabbitMQ · boto3 · docker compose
- **AI** — OmniRoute (LLM gateway) · OpenRouter (vision/image) · DeepSeek/opencode-go · GLM-5v · Basic Pitch · MuScriptor · piper TTS
- **AI** — OmniRoute (LLM gateway) · OpenRouter (vision/image) · DeepSeek/opencode-go · GLM-5v · Basic Pitch · MuScriptor · piper TTS · **faster-whisper** (voice_whisper on .13:5000, shared, speech-to-text)
- **Apps/services** — Garage S3 · Caddy · Gitea · Vikunja · Outline · Home Assistant · Metube · Jellyfin · Nextcloud · Immich (+ more)
- **Infra** — NixOS (.13/.27) · Ubuntu (.35) · pi agent · systemd · RustDesk · Borg