Powerful, private, and efficient speech-to-text using your device's WebGPU capabilities.
On-Device Processing
Uses your hardware's WebGPU for fast, private transcription
100+ Languages
Supports transcription across 100 different languages
Real-Time Feedback
Instant transcription results as audio is processed
Technologies Used
by this browser :(
Efficient Model Caching
Download the model once, and it's stored for future visits.
Reduces bandwidth usage and provides instant startup on return visits.
Multilingual Support
Accurately transcribe audio in over 100 different languages.
From English to Japanese, Arabic to Spanish - global language coverage.
Real-Time Feedback
See transcription results as the audio is being processed.
Immediate results with continuous updates.
Web Worker Technology
Processing happens in separate threads, keeping the UI responsive.
Offloaded processing ensures smooth user experience.
Complete Privacy
You are loading whisper-base, a 73 million parameter speech recognition model. Once downloaded, the model (~200 MB) will be cached and reused when you revisit the page.
With on-device processing using WebGPU, your audio never leaves your computer. The entire transcription happens locally, ensuring maximum privacy and security for sensitive content.