AI-PoweredMeeting NotesThat Stay Private

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

Wavezard captures system and microphone audio directly and transcribes it locally using selectable Whisper models (you can choose different models for speed vs. accuracy). Speakers are auto-identified and persist across meetings. "Ask Your Meeting" enables local chat with the meeting in context. 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 join your calls.

Yes. Wavezard supports as many languages as the local Whisper models do. Whisper currently 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. Speech detected in a non-English language is automatically converted to English.

An internet connection is required to activate the software and for periodic license checks. Other than that, no internet connection is needed for core functionality, i.e. recording, transcription, 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 Wavezard to external platforms (Slack, Discord, Zapier, etc.); these optional features also require internet.

Wavezard 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 faster. Wavezard ships as a portable single binary executable, meaning there is 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 machine learning tasks such as the voice activity detector, noise cancellation, and speaker identification are lightweight. Choose smaller Whisper models for lower resource use.

Wavezard uses Silero VAD for voice activity detection, RNNoise for noise cancellation, Whisper for speech recognition, Resemblyzer for speaker identification, and Qwen3 for LLM chat.

Download the app, launch it, and start capturing a meeting!

Pricing is simple: you can purchase 1 license for $19 (plus applicable taxes) to activate 1 device and use the software in perpetuity. A 14-day free trial is available for you to try before making a purchase decision.