Back to Home
COBACOBAINI icon

vibe

by COBACOBAINI

Overview

A fast, cross-platform desktop application for transcribing audio locally using Whisper models, with optional integration for OpenAI/Claude for summarization.

Installation

Run Command
vibe --model <path_to_model.bin> --file <path_to_audio.wav> [--language <language>] [--temperature <temp>] [--n-threads <threads>] [--translate] [--max-text-ctx <tokens>] [--init-prompt <prompt>] [--write <output_path>] [--format <format>] [--word-timestamps] [--max-sentence-len <len>] [--diarize] [--diarize-vad-model <path>] [--diarize-speaker-id-model <path>]

Environment Variables

  • RUST_LOG

Security Notes

The application executes external binaries (FFmpeg, yt-dlp) and modifies system configurations (Windows Registry, apt packages on Linux). The `scripts/pre_build.js` downloads these external dependencies, which introduces supply chain risks if the download URLs or the binaries themselves are compromised. A critical concern is the `custom_command` option for FFmpeg, which could allow arbitrary command injection if user input is not properly sanitized. The `tauri-plugin-http` uses `unsafe-headers` for Ollama communication, potentially allowing arbitrary `Origin` headers. Screen capture on macOS also requires specific permissions.

Similar Servers

Stats

Interest Score34
Security Score5
Cost ClassLow
Stars4
Forks0
Last Update2025-12-06

Tags

aicross-platformdesktopopenairust