# Whisper Web — Full Documentation > Whisper Web is a browser-first speech-to-text and media utility suite powered by OpenAI's Whisper AI and FFmpeg WebAssembly. Free local transcription and media conversion run on the user's device; Whisper Web Unlimited also offers an optional secure cloud workflow for larger files. ## Available Languages on Website The full website UI is available in the following 19 languages on https://whisperweb.dev: - English (English) (en) — / — Content included below - German (Deutsch) (de) — /de/ — Visit website for content - French (Français) (fr) — /fr/ — Visit website for content - Japanese (日本語) (ja) — /ja/ — Visit website for content - Korean (한국어) (ko) — /ko/ — Visit website for content - Arabic (العربية) (ar) — /ar/ — Visit website for content - Portuguese (Brazil) (Português (Brasil)) (pt-BR) — /pt-BR/ — Visit website for content - Spanish (Español) (es) — /es/ — Visit website for content - Italian (Italiano) (it) — /it/ — Visit website for content - Vietnamese (Tiếng Việt) (vi) — /vi/ — Visit website for content - Dutch (Nederlands) (nl) — /nl/ — Visit website for content - Polish (Polski) (pl) — /pl/ — Visit website for content - Greek (Ελληνικά) (el) — /el/ — Visit website for content - Hebrew (עברית) (he) — /he/ — Visit website for content - Swedish (Svenska) (sv) — /sv/ — Visit website for content - Norwegian (Norsk) (no) — /no/ — Visit website for content - Danish (Dansk) (da) — /da/ — Visit website for content - Finnish (Suomi) (fi) — /fi/ — Visit website for content - Hungarian (Magyar) (hu) — /hu/ — Visit website for content --- ## Product Overview Whisper Web brings OpenAI's Whisper speech recognition model and private media conversion directly into the user's web browser. Free local transcription uses WebAssembly and WebGPU, while browser media conversion uses FFmpeg WebAssembly. Users who explicitly choose Whisper Web Unlimited can use a secure cloud workflow for larger files and account features. Key value propositions: - **Private Local Tools**: Free local transcription and `/convert/` media tools process files in the browser without uploading the media. - **Currently Free Local Mode**: Local processing runs on the user's own hardware. This pricing model may evolve as the product develops. - **99 Spoken Languages**: Supports the 99 language choices exposed by the Whisper model in the product, including English, Chinese, Spanish, French, German, Japanese, Korean, and Hindi. - **No Account Required**: Users can start transcribing immediately without sign-up or registration. - **Offline Capable**: Once the AI model is cached in the browser, transcription works without an internet connection. ## How It Works ### Architecture Whisper Web uses a browser-first hybrid architecture: 1. The user visits https://whisperweb.dev in a modern browser. 2. The Whisper AI model (~75 MB for the Base model) is downloaded from Hugging Face's CDN and cached in the browser. 3. In Free local mode, audio is processed in-browser using WebAssembly (via ONNX Runtime) and optionally WebGPU for hardware acceleration. 4. Browser media converters run FFmpeg WebAssembly locally; converter source files are not uploaded. 5. Users may explicitly choose Whisper Web Unlimited for secure cloud processing of larger files, batch upload, and cross-device sync. 6. Transcription and conversion results can be downloaded or exported. ### Supported Input Methods 1. **File Upload**: Drag and drop or select audio/video files (.mp3, .wav, .m4a, .mp4, etc.) 2. **Microphone Recording**: Record audio directly from the browser using the device microphone. 3. **Direct Media URL**: Load a direct public audio or video file URL that the browser can fetch. Ordinary YouTube or Bilibili watch-page URLs are not supported importers. ### Export Formats - **TXT**: Plain text format, ideal for notes, articles, and documentation. - **JSON**: Timestamped chunks with start/end times, suitable for programmatic use. - **SRT (SubRip Subtitle)**: Industry-standard subtitle format compatible with YouTube, Premiere Pro, VLC, and most video editors. - **VTT (WebVTT)**: HTML5-native subtitle format for web video players and HTML5 `