Whisperstream

Whisperstream is a Windows dictation tool that runs entirely on your PC. You press a key, speak, and your words appear as text wherever your cursor is — in Gmail, Slack, VS Code, or any other app. The whole thing happens on your machine, using a local AI model. No audio ever leaves your computer, and you don't need an account or an internet connection after setup.

The tool uses OpenAI's Whisper model (NVIDIA-optimized) to transcribe speech locally. It includes an AI cleanup feature that removes filler words like "um" and "uh" and applies spoken corrections — for example, if you say "Friday, sorry, Saturday" it outputs just "Saturday." You can build a custom dictionary for names, jargon, and acronyms (so "next js" becomes "Next.js"). It supports 25 languages, all running locally. There's also a transcript history with search and audio replay, the ability to import audio files for transcription, and per-app custom AI prompts. The installer is code-signed with a Microsoft-trusted certificate.

Whisperstream is ideal for anyone who types a lot and wants to go faster — developers, writers, journalists, and professionals who spend their days in email, documents, or code. It's also a strong choice for privacy-conscious users who don't want their voice data sent to a cloud server. The one-time $29 price (for up to 2 PCs) makes it a cheap alternative to subscription-based dictation services.