Speech to Text
Transcribe or translate speech from audio and video files, or from your microphone, with Whisper.
Runs on your device — your files are never uploaded.
Drop an audio or video file here or click to choose
MP3, WAV, M4A, OGG, WebM, MP4 and other formats your browser can play.
Your browser will ask for microphone permission. The recording stays in this tab.
Input
Transcript
Model: OpenAI Whisper (MIT licence), ONNX conversion by onnx-community, run with Transformers.js (Apache-2.0) in a background worker. The model is downloaded once from Hugging Face and cached by your browser. Uses WebGPU when your browser supports it, otherwise WebAssembly. Audio is converted to 16 kHz mono locally; long recordings are processed in 30-second chunks.