Soundmind

Soundmind is an Ubuntu desktop assistant for system audio. It captures the current output monitor, transcribes speech in near real time, detects likely questions, and lets you generate answers, summaries, and commentary while keeping session history and agent setup locally in SQLite.

Current release: v0.2.1

What it does

• Captures desktop audio through PulseAudio or PipeWire's Pulse compatibility layer
• Streams audio to ElevenLabs Scribe Realtime for transcription
• Uses OpenAI for manual answer, summary, and commentary actions
• Shows live transcript and assistant output in a Tauri desktop UI
• Flags detected questions explicitly in the UI
• Lets you configure a default interview-assistant instruction
• Lets you upload priming documents such as your CV, job description, or notes
• Stores sessions, transcript segments, settings, priming documents, and assistant output locally
• Supports tray controls, global shortcuts, history, export, and systemd --user installation

Quick start

Install from Debian package

Download the latest soundmind_*_amd64.deb from GitHub Releases, then:

sudo apt install ./soundmind_0.2.1_amd64.deb
soundmind-setup-user
soundmind

The package installs the desktop app, backend, config example, desktop entry, icon, and a systemd --user service unit. soundmind-setup-user prepares ~/.config/soundmind and enables the backend for the current user.

Run from source

1. Copy the config:

cp config.example.toml config.toml

2. Put your provider keys in keys.env:

OPENAI_API_KEY=...
ELEVENLABS_API_KEY=...

3. Start the backend:

cargo run -p app_backend

4. Start the desktop UI:

cargo run -p app_ui

5. In the desktop UI, open Agent Configuration to:

• edit the default interview-assistant instruction
• upload priming documents for the model to use

Best results come from text or markdown files. PDF upload also works when pdftotext is installed on the machine.

6. If you want the terminal debug client instead:

cargo run -p app_ui --bin terminal_ui

Install as a user service

./scripts/install-user-service.sh
soundmind

This installs release binaries under ~/.local, writes config to ~/.config/soundmind, and enables soundmind-backend.service with systemctl --user.

Default shortcuts

• Ctrl+Alt+Shift+M: show or hide the main window
• Ctrl+Alt+Shift+A: answer the detected or last question
• Ctrl+Alt+Shift+S: summarise the last minute
• Ctrl+Alt+Shift+C: comment on the current topic

If another app already owns one of these shortcuts, Soundmind still launches.

Documentation

• Installation guide
• User guide
• Development guide
• Architecture notes
• Implementation roadmap

Current status

The current build includes:

• real system-audio capture
• real ElevenLabs realtime STT integration
• OpenAI-backed manual answer, summary, and commentary actions
• question detection surfaced in the desktop UI
• local settings, history, export, and privacy status
• configurable interview instruction plus uploaded priming documents
• tray integration and best-effort global shortcuts
• packaging helpers and systemd --user assets
• Debian release packaging and GitHub release tooling