Model and license credits

MadLadPad runs speech recognition in your browser. The first transcription downloads model and runtime files from their maintainers; your selected recording is processed on your device.

• Speech model: Xenova/whisper-tiny.en, an ONNX conversion of OpenAI Whisper Tiny English. The Xenova model is marked Apache-2.0; the original Whisper project is MIT licensed. Read the Apache-2.0 license and Whisper MIT license.
• Inference library: Hugging Face Transformers.js, Apache-2.0. Read the Apache-2.0 license.
• Audio inference runtime: ONNX Runtime Web, MIT. Read the ONNX Runtime MIT license.

Model and runtime downloads are separate from transcription: the selected audio is not uploaded by this feature. Recognition is English-only; review its editable words and timestamps.