Skip to content

onepunk/notely-ai

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Notely AI

Standalone AI desktop client. An Electron app that performs meeting transcription and AI-powered summarization locally on the user's machine, without sending audio or transcripts to a cloud service. Fully standalone — no backend, no license server, no activation.

What it does

  • Records meetings from microphone + system audio
  • Transcribes speech to text using a local Whisper model (CPU or GPU)
  • Generates meeting summaries using a local LLM via llama.cpp
  • Stores everything in a local encrypted SQLite database
  • Runs entirely on your machine with no cloud dependency

Every feature is unlocked by default. There is no license, no activation step, no upgrade prompt.

Prerequisites

  • Node.js 20+
  • Python 3.11+ (for the local transcription pipeline)
  • A C++ toolchain (for the native addons under native/)
  • OpenSSL development headers
  • Platform-specific build tools:
    • Ubuntu/Debian: build-essential g++ make python3-dev libsecret-1-dev libssl-dev
    • macOS: Xcode command-line tools (OpenSSL via Homebrew)
    • Windows: Visual Studio Build Tools with the "Desktop development with C++" workload; OpenSSL for Windows

Development setup

  1. Install Node.js dependencies:

    npm install
  2. Build the native addon, main bundle, renderer, and preload:

    npm run build
  3. Set up the Python transcription environment:

    bash scripts/setup-python-env.sh
  4. Download a Whisper model into src/main/transcription/models/ (e.g., tiny.en.pt for quick CPU testing, or a larger model for better accuracy).

  5. Start the dev app:

    npm run dev

Development-only environment variables

  • DEBUG_DB=true — disables SQLCipher database encryption. Development only. The app logs a prominent warning at startup when it is active.
  • NOTELY_MODEL_DIR — override the default Whisper model directory
  • NOTELY_MODEL_NAME — choose which .pt model file to load
  • NOTELY_USE_GPU — enable CUDA acceleration if a compatible GPU is present

Building a release

The public repo ships a minimal CI workflow (.github/workflows/ci.yml) that runs lint and type-check only. It does not build signed release artifacts.

If you want to produce signed builds of your own, fork the repo and add a workflow that configures the relevant signing credentials as GitHub Actions secrets (CSC_LINK, CSC_KEY_PASSWORD, APPLE_ID, APPLE_APP_SPECIFIC_PASSWORD, APPLE_TEAM_ID, etc.) and calls electron-builder.

Related repositories

  • notely-cloud — cloud desktop client that syncs notes with a self-hosted backend
  • notely-platform — microservices backend powering notely-cloud

Security

To report a vulnerability, see SECURITY.md.

License

Apache License 2.0. See LICENSE.

About

Standalone AI desktop client — Electron app with local Whisper transcription and llama.cpp summarization. Runs entirely on your machine, no backend or license server required.

Topics

Resources

License

Contributing

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors