Office Recorder

Local, always-on office audio capture with end-of-day transcription and summaries. Designed for Mac Studio workstations and on-prem processing.

Mac Studio Setup

Follow MAC_STUDIO_SETUP.md for step-by-step install and launchd auto-start.

Features

• One-click recording for the whole day (chunked audio segments).
• Local transcription (faster-whisper) and conversation batching.
• Daily summary with topics, decisions, and action items.
• Local web UI plus CLI control script.
• Optional OpenClaw webhook for sending the daily overview.
• Optional auto schedule (start/stop by time window).
• Optional diarization (speaker separation) module.

Quick Start (Mac Studio)

1) Install dependencies

brew install ffmpeg

cd office-recorder/backend
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

2) Configure audio + LLM

Copy the environment template:

cp .env.example .env

List audio devices (Mac):

ffmpeg -f avfoundation -list_devices true -i ""

Set OFFICE_RECORDER_AUDIO_INPUT in .env (e.g., :0 or :Scarlett 2i2).

For local LLM, install and run Ollama or LM Studio. Example with Ollama:

ollama pull llama3.1:70b

Update .env:

OFFICE_RECORDER_LLM_BASE_URL=http://localhost:11434
OFFICE_RECORDER_LLM_MODEL=llama3.1:70b

3) Run the server

python -m office_recorder

Open the UI:

http://127.0.0.1:8787

Control Script

Use the control script for quick CLI actions:

./scripts/office_recorder_ctl.sh start
./scripts/office_recorder_ctl.sh stop
./scripts/office_recorder_ctl.sh status
./scripts/office_recorder_ctl.sh transcribe
./scripts/office_recorder_ctl.sh summarize

Make it executable on macOS:

chmod +x ../scripts/office_recorder_ctl.sh

OpenClaw Integration

Use the skills/office-recorder/SKILL.md skill. The skill triggers the control script via OpenClaw Exec. Configure Exec to allow the script path and run on the Mac Studio host.

Data Layout

All data is stored in ~/OfficeRecorder by default:

OfficeRecorder/
  2026-01-30/
    audio/
    transcripts/
    summaries/
    session.json

Tests

pip install -r requirements-dev.txt
pytest