AI-PoweredMeeting NotesThat Stay Private

Notecove is an AI notetaker that records, transcribes, diarizes, summarizes, and lets you chat with your meeting content, all entirely on your device. No bots join your calls.

Notecove captures system and microphone audio directly and transcribes it locally using selectable Whisper models (you can choose different sizes for speed vs. accuracy). Speakers are auto-identified and persist across meetings. "Ask Your Meeting" enables local RAG-style chat. Use customizable templates for summaries, Bookmarks to mark moments, and an integrated Notepad to jot down thoughts during the meeting. Webhooks let you share to external platforms.

Anywhere audio plays on your computer: Zoom, Microsoft Teams, Google Meet, Webex, Discord, WhatsApp, or in-person sessions. Because it records at the OS level, there are no integrations to set up and no bots joining calls.

Yes. Notecove supports as many languages as the local Whisper models do. Currently, Whisper claims to support 99 languages. Smaller models may not perform very well on non-English, so you may need a bigger model for multilingual meetings. The bigger the model, the more processing power it needs. You can pick model sizes to balance quality, speed, and accuracy. It is also recommended to turn on noise cancellation even if you are not in a noisy environment, as it helps Whisper recognize speech better.

An internet connection is required to activate the software for the first time. Thereafter, no internet connection is needed for recording, transcription, search, summary, or chat, and everything runs on your device. If you choose to use an optional external LLM via API, an internet connection is required to call that API. You can also enable optional webhook integrations that send data from Notecove to external platforms (Slack, Discord, Telegram, Zapier, etc.); these optional features require the internet.

Notecove is designed for everyone, and not everyone has a high-performance computer. It runs well on a modern CPU. If you have a GPU, it will use it and run even faster. Notecove ships as a portable single binary executable, meaning there's no need to install anything else; no Docker, Node, or Python required. It runs on macOS 14.2+ and Windows 10/11. It uses Metal on macOS, Vulkan on Windows, and has an efficient CPU fallback. Most features such as noise cancellation, speaker identification, and vector embeddings are lightweight. Choose smaller Whisper models for lower resource use.

Notecove uses RNNoise for noise cancellation, Silero VAD for voice activity detection, Whisper for speech recognition, WeSpeaker for speaker identification, Snowflake Arctic Embed for text embeddings, and Qwen3 for LLM chat.

Download and launch the app, set up audio sources, select the models to use, and start recording!

Pricing is simple: you can purchase 1 license for $19 (plus applicable taxes) to activate 1 device and use the software in perpetuity. A 30-day free trial is available.