orukAPI documentation ↗

Speech,
in words.

Try Qwen3-ASR, served by Oruk.

Upload a short audio file and get its transcript. The model detects the language automatically.

10 secmaximum duration2 MiBmaximum file size

A shared free demo: 100 attempts per day, one request at a time. Resets at 00:00 UTC.

No file selected.

Audio is sent to Oruk for processing. Oruk does not store or log demo audio or transcripts. Invalid files and busy requests also count toward the shared daily budget.

Qwen3-ASR-1.7B is developed by Qwen, under Apache 2.0. Hosted by Oruk. This community Space is separate from native Hugging Face Inference Provider approval.