Convert audio and video to text in 99 languages — AI runs in your browser
Turn any audio or video file into accurate written text, completely free and without creating an account. Our online audio transcriber uses OpenAI's Whisper model — a state-of-the-art speech recognition system — running directly inside your web browser. Upload an MP3, WAV, M4A, OGG, FLAC, MP4, WEBM or MOV file, pick a language (or let the tool detect it automatically), and get a clean transcript in seconds. Export the result as plain text (.txt) or as timed subtitles in .srt and .vtt format, ready to use on YouTube, Vimeo, or any video editor.
Unlike most online transcription services, this tool never uploads your files to any server. The AI model is downloaded once into your browser cache (approximately 75 MB for the fast model, 150 MB for the balanced one) and from that point on the transcription happens entirely offline on your own device. This means zero privacy concerns, no subscription fees, no upload limits, and no throttling — ideal for journalists, podcasters, students, researchers, lawyers, and anyone who needs to transcribe sensitive recordings without exposing them to third parties.
The transcriber supports 99 languages, including Italian, English, Spanish, French, German, Portuguese, Dutch, Russian, Chinese, Japanese, Korean, Arabic, Hindi, Turkish, and Polish. Automatic language detection works well for clean recordings; for noisy audio or strong accents, selecting the language manually typically improves accuracy. Two model sizes are available: the "tiny" model is very fast and ideal for short recordings or lower-powered devices, while the "base" model offers noticeably better accuracy on longer content and difficult audio at the cost of a larger initial download and slower processing.
Typical use cases include transcribing interviews, podcasts, lectures, meetings, voice notes, WhatsApp vocal messages, YouTube videos, webinars, court hearings, and any recorded conversation you want to search, edit, or translate. Because the output preserves word-level timestamps, you can also generate synchronized subtitles for your videos in one click. For best results, use recordings with clear speech and minimal background noise; files under 30 minutes process fastest, but longer files up to about an hour work fine on most modern browsers. The tool is powered by open-source technology (Whisper, MIT license; Transformers.js, Apache 2.0) and runs 100% in your browser — no registration, no upload, no fuss.