hjLabs AI Playground
Starting the on-device runtime…
Drop in a file or record from your microphone and Whisper transcribes it on your GPU. whisper-tiny.en is 40 MB and near-instant; whisper-base and whisper-small trade speed for accuracy and handle accents, noise and 99 languages. Because nothing is uploaded there is no file-size ceiling, no per-minute pricing, and no confidentiality problem transcribing an interview, a therapy session or a board meeting. Timestamps and a plain-text export come out the other side.
free speech to textwhisper online freebrowser transcriptiontranscribe audio privatelyoffline speech recognitionwebgpu whisperunlimited transcription freeaudio to text no upload
No hard limit. Whisper processes 30-second windows internally and the page chunks longer recordings automatically. A one-hour file is fine; it just takes proportionally longer.
tiny.en for quick English notes, base for a good general default, small when accuracy matters more than speed. Anything above small is impractical in a browser today.
No. The model weights are downloaded from Hugging Face to your browser once, and every inference after that runs on your own GPU through WebGPU. Your text, images and audio are never sent anywhere — hjLabs.in has no inference server and no way to see your input.
Small models do. Recent mobile Chrome and Safari expose WebGPU, but memory is the binding limit — stay under roughly 200 MB of weights on Android, and expect iOS to reclaim the tab sooner. The catalog shows every download size up front so you can pick something that fits.
That depends on the individual model, not on this site. Each model page links to its Hugging Face repository, where the licence is stated — most of the catalog is Apache-2.0 or MIT, a few are research-only. We add nothing on top: hjLabs.in claims no rights over anything you produce here.