Transcription has long meant paying per minute or spending hours typing, and uploading sensitive recordings to servers you do not control. Transcribela takes a different approach. Open-source speech recognition (OpenAI’s Whisper) and modern browser technology (WebGPU and WebAssembly, through Hugging Face Transformers.js) make it possible to run the model on your own device.
That gives three things at once: it is free, because we pay nothing per minute; it is private, because your recordings are never uploaded; and it is unlimited, because your computer does the work.
How we keep content accurate
Guides and reference pages are checked against primary sources such as official documentation, standards and regulations, and each page shows when it was last updated. Read our editorial policy.
Credits
Speech recognition by OpenAI Whisper (MIT license). In-browser inference by Hugging Face Transformers.js and ONNX Runtime Web. Models converted and hosted by the Hugging Face community.
Contact
Questions, ideas or corrections: contact us.