How Audio Transcription Works
AnyTool transcribes speech entirely in the browser using a self-hosted Whisper automatic speech recognition (ASR) model run via transformers.js on WebAssembly/WebGPU. The audio is decoded with the Web Audio API, resampled to the model rate, and processed locally to produce text with word/segment timings. The model weights download once and are cached, after which transcription works offline. Output can be exported as plain TXT or as timestamped SRT and VTT subtitle files. No audio or text is ever uploaded and no CDN inference is used.
- Runs client-side — audio is never uploaded
- Self-hosted Whisper ASR via transformers.js (WASM/WebGPU)
- Model is cached, so it works offline after first load
- Segment timings drive SRT/VTT subtitle export
- Export TXT, SRT and VTT
- No per-minute cap, account or CDN inference
