Set up the Vibe desktop app for offline Whisper transcription on Mac, Windows and Linux, and fix model download errors, GPU detection and batch issues.
Download from the Vibe site or GitHub releases (Tauri app, all three platforms).
First launch downloads a Whisper model; you can switch sizes in Settings → Models.
Drop multiple files at once — Vibe queues them; exports per-file to TXT/SRT/VTT/JSON.
Same Whisper-class accuracy, no install, no model downloads, no GPU questions. Speaker labels, word timestamps and every export (TXT, SRT, VTT, DOCX) included. Free instant preview; credits from $2 and they never expire.
The models come from Hugging Face; if blocked, download the GGML .bin manually and point Settings → Models at the file.
Vibe uses Vulkan for GPU on Windows — update GPU drivers; on old iGPUs it silently falls back to CPU.
Auto-detect reads the first 30 seconds — if the intro is music or another language, set the language explicitly.
No — it's file-only. Use a hosted service that fetches URLs server-side, then keep Vibe for private local files.
More on Vibe: the full Vibe page · Vibe alternatives · all transcription tools