Show HN: VocalSplit – Browser-based vocal remover running 100% locally

Posted by jacoka 7 hours ago

I built VocalSplit because I wanted a way to separate vocals and music without uploading files to a remote server. Most existing tools are either paid SaaS or require a complex local Python environment (like Spleeter or Demucs).

This tool runs entirely in the browser. No data leaves your machine, and there are no processing limits or queues. It uses [említsd meg a technológiát, pl. WebAssembly vagy ONNX Runtime Web] to handle the inference locally.

Thanks for checking it out!

Comments

Comment by jacoka 7 hours ago

The app uses Spleeter models running in an ONNX environment (via ONNX Runtime Web).

I wanted to eliminate the need for server-side processing and Python dependencies. By porting the inference to the browser, everything stays 100% local, which is great for privacy and means no more waiting in queues for 'free' online tools.

It's still being optimized for different hardware, but I'm curious to hear how it performs on your machines!