Whishper is a web-based audio transcription and subtitle editing suite that runs entirely on your local machine.
The tool solves the problem of transcribing audio and video to text while maintaining complete privacy and offline capability. It uses Faster-Whisper as its transcription backend, allowing users to process media files or URLs through a web interface without sending data to external services. The architecture consists of a transcription API that runs Faster-Whisper models, a backend service coordinating requests and database operations, and a frontend web UI built with Svelte and SvelteKit.
Whishper suits users who prioritize privacy, need offline transcription capability, or want to avoid recurring API costs from cloud-based services. It supports GPU acceleration for NVIDIA hardware and CPU-only operation, making it accessible across different machine configurations. The tool goes beyond basic transcription by including a subtitle editor with features like segment splitting, insertion, and character-per-second warnings, plus the ability to translate transcriptions using Libretranslate. Users can export results in multiple formats including TXT, JSON, VTT, and SRT.
The project is undergoing a complete rewrite on a separate branch, and the main branch is not receiving new releases or updates during this transition. Contributions are actively welcomed through pull requests and issue resolution. Development documentation is available for those interested in setting up a local development environment.
