Vibe is a new open-source tool for multilingual audio transcription that will make you excited! Gone are the days when you had to settle for approximate subtitles or wait forever to get a decent transcription.
To achieve this, it uses Whisper AI, developed by the geniuses at OpenAI, which I have mentioned to you many times. This cutting-edge speech recognition model is capable of transcribing an astounding number of languages with impressive accuracy, making Vibe a truly versatile audio solution packed with features.
For instance, you can transcribe audio and video files in batches, preview the results in real-time, export in a multitude of formats (SRT, VTT, TXT…), and even customize the models according to your needs. It works entirely offline, so there’s no risk of your sensitive data falling into the hands of big tech companies, and it runs on macOS, Windows, and Linux. Simply visit the GitHub releases page and download the version that corresponds to your OS.
Support for Apple Silicon is optimized, offering increased performance. For Windows, version 8 or higher is required, but I assume most of you are already on Windows 10/11. Linux users can install Vibe via a .deb file, and Arch Linux users can use debtap to convert the package to suit their needs.
In terms of performance, it’s a piece of cake. As you might have guessed, Mac computers benefit from a little GPU optimization that boosts results. But even on an old Windows machine, Vibe can adapt to your resources without complaint thanks to its advanced settings. And for Linux users, note that support for system audio and microphone input is coming soon.
In short, it’s worth a try if you’re in the subtitle or transcription business.
