WaveFlow is a private voice dictation app for Windows and macOS. Instead of sending your voice to a cloud service, WaveFlow runs a powerful speech-to-text engine called NVIDIA Parakeet directly on your own PC, home server, or VPS. This means:
Think of it as having a super-smart assistant who lives inside your computer and never reports back to anyone.
Follow these simple steps to download and start using WaveFlow today.
Visit this link to download the application.
Once the file finishes downloading, locate it in your Downloads folder. Double-click the file to begin installation. Follow the on-screen prompts – it’s just like installing any other program. No special knowledge required.
After installation, find the WaveFlow icon on your desktop or in your Start Menu. Click it to open the app.
Ctrl + Shift + Space) to start listening.WaveFlow runs a local speech recognition engine called Parakeet. This is a Neural Network developed by NVIDIA that is incredibly accurate at understanding human speech. The engine runs entirely on your hardware, which is why everything stays private.
WaveFlow uses a user-friendly interface built with PySide6 (a modern Python toolkit), so the app feels smooth and responsive even on modest hardware.
WaveFlow is designed to run on equipment most people already own. Here are the typical recommendations:
| Component | Minimum | Recommended |
|---|---|---|
| Processor | 4-core CPU | 8-core or better |
| Memory (RAM) | 8 GB | 16 GB |
| Storage | 2 GB free space | 5 GB free space |
| Graphics | Integrated GPU | NVIDIA GPU (for best performance) |
| Microphone | Any built-in mic | USB headset or desktop mic |
💡 Tip: If you have a newer NVIDIA graphics card, WaveFlow can use it to boost speed and accuracy even further.
If you want even more power or need to share WaveFlow with family or teammates, you can run it as a Docker container on a home server or virtual private server (VPS).
This is especially useful if you have an older PC that you want to leave always on, or if you prefer a centralized setup. The Docker image includes everything needed – just start the container, and connect from your Windows or macOS computer.
Yes, 100%. WaveFlow processes your speech entirely on the machine where it runs. There is no cloud service, no telemetry, no account, and no network calls to external services. It works even with your internet connection completely disabled.
Absolutely. That’s one of its core features. WaveFlow works perfectly offline.
WaveFlow currently supports English (US, UK, and other regional accents) with high accuracy. Additional languages are planned based on community demand.
Yes, WaveFlow is fully open-source and free to use. No subscriptions, no hidden fees, no premium tiers.
Almost certainly yes. If your computer can record audio (which all modern computers can), WaveFlow will detect and use your microphone.
Yes. WaveFlow types into whatever text field is currently active – Word, Notepad, browsers, email clients, and more.
Try downloading the latest version from the download page. Make sure your system meets the minimum requirements listed above.
WaveFlow offers several settings so you can make it work exactly how you like:
WaveFlow checks for updates automatically on startup. When a new version is available, you’ll see a download link directly in the app, or you can always grab the latest from the releases page. Updating preserves all your settings and custom vocabulary.
1. Total Privacy
Your thoughts are yours. WaveFlow respects that.
2. Zero Subscription
Pay nothing. Use it forever. No surprises.
3. Works Perfectly Offline
No internet? No problem.
4. Accurate & Responsive
NVIDIA’s Parakeet engine is among the best speech recognition models in the world.
5. Easy for Everyone
No command lines, no coding, no jargon. Download, install, dictate.
Don’t wait – privacy-respecting dictation is one click away.
Visit this link to download the application.
WaveFlow – the voice assistant that belongs to you.
Keywords: dictation, docker, local-first, nvidia-nemo, offline, onnx, parakeet, privacy, pyside6, python, self-hosted, speech-to-text, stt, voice-typing, windows