Vibe

Vibe is a free, open-source desktop app that runs OpenAI's Whisper model locally via whisper.cpp to transcribe audio and video without sending anything to the cloud. Built on Tauri, it ships native binaries for macOS, Windows, and Linux, supports GPU acceleration across all three, and handles dozens of languages with optional translation to English. Output formats include SRT, VTT, TXT, HTML, PDF, JSON, and DOCX, which is wider coverage than most local Whisper wrappers bother with.

Where Vibe earns its place is the feature density for a hobbyist project: speaker diarization, microphone and system-audio capture, batch transcription of folders, a CLI, and a local HTTP API for scripting. There's also an optional Claude integration for AI summarization on top of the transcript. It's aimed at content creators, journalists, researchers, and privacy-conscious users who don't want to upload sensitive recordings to a SaaS. Because it's MIT-licensed and entirely local, there's no pricing, no quota, and no account.

The trade-offs are the usual ones for local Whisper: transcription speed depends on your GPU, and the largest models need real VRAM to be tolerable. iOS and Android are listed as coming soon, so if you need mobile capture today, look elsewhere. There's no managed cloud option, so collaborative or team workflows have to be built on top of the HTTP API.