hjLabs AI Playground
Starting the on-device runtime…
Audio classification answers "what is this noise?" rather than "what words were said". Audio Spectrogram Transformer models tagged on AudioSet recognise hundreds of everyday sound events — speech, music, engines, alarms, animals — while Wav2Vec2 keyword-spotting models handle short spoken commands. Record or upload a clip and you get ranked labels with confidence scores, computed locally.
audio classification onlinesound recognition aiaudioset classifierkeyword spotting demowav2vec2 browserclassify audio freeaudio tagging tool
No. These models recognise categories of sound, not individual recordings — they will tell you "music, guitar, singing", not the track title. Audio fingerprinting is a different technology.
Anything your browser can decode: WAV, MP3, M4A, OGG, FLAC. The page resamples to the 16 kHz mono the models expect.
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.