Upload a file and transcribe in one step — models cache in OPFS after the first run.
For manual downloadModel() or live mic flows, see all demos .
Runs on WebGPU when available (Chrome, Edge, other Chromium browsers), otherwise falls back to WebAssembly (slower). Optional Linux tip: Chrome may need --enable-unsafe-webgpu --enable-features=Vulkan.
✕
📂
Choose file
Moonshine Tiny (~55 MB)
Moonshine Base (~98 MB)
Parakeet CTC 0.6B (~612 MB)
Parakeet TDT 0.6B v3 (~1.26 GB)
Whisper Tiny (~120 MB)
Whisper Base (~206 MB)
Whisper Small (~586 MB)
Whisper Tiny Timestamped (~120 MB)
Whisper Base Timestamped (~206 MB)
Whisper Small Timestamped (~586 MB)
Lite Large V3 Turbo Fast (~1.59 GB)
Lite Large V3 Turbo (~1.84 GB)
Lite Large V3 Turbo Acc (~2.02 GB)
Whisper Large V3 Turbo (~2.88 GB)
Whisper Large V3 Turbo Timestamped (~2.88 GB)
Distil Whisper Small (~538 MB)
Transcribe
Model credits: Parakeet TDT 0.6B v3 by NVIDIA, licensed under CC BY 4.0 . Changes: converted to ONNX and quantized by istupakov /ysdede ; runs with parakeet.js (MIT). Parakeet CTC 0.6B by NVIDIA, licensed under CC BY 4.0 . Changes: converted to ONNX and quantized by onnx-community . Whisper by OpenAI (MIT). Moonshine by Useful Sensors (MIT). Distil-Whisper by Hugging Face (MIT). Lite-Whisper by efficient-speech (Apache-2.0). ONNX exports by onnx-community . All third-party notices .
Timestamped model — click words while playing to seek; words highlight in sync.
Ready — select an audio or video file to begin.
🎙️
Transcript segments will appear here as they are transcribed.