'%20/%3e%3c!--%20Checkmark%20--%3e%3cpath%20d='M10%2016L14%2020L22%2012'%20stroke='white'%20stroke-width='3'%20stroke-linecap='round'%20stroke-linejoin='round'/%3e%3c/svg%3e)
Audio/Video Transcription
Convert speech to text with 100% privacy using Whisper. This offline transcription tool runs in your browser, with audio and video support.
Transcribe podcasts, meetings, interviews, lectures, or any audio/video content. Perfect for creating captions, notes, or text records while maintaining complete privacy.
How to use Audio/Video Transcription
- Upload an audio or video file
- Select the Whisper model (Tiny English)
- Click "Transcribe Audio" to start processing
- View the transcript with timestamps
- Copy or download in TXT, SRT, or VTT format
Speech recognition that runs on your device
Transcription uses a Whisper speech-recognition model that downloads to your browser and runs locally. The audio never leaves your machine, which is the entire point for recordings of meetings, interviews, medical appointments, or anything else you would not upload to a third party.
The first run downloads model weights, which takes time and bandwidth depending on the size you pick. After that the model is cached and subsequent transcriptions start immediately.
Choosing a model size
Smaller models start faster and use less memory; larger ones are more accurate, particularly on accented speech, background noise, and technical vocabulary. Tiny is a reasonable starting point for clear single-speaker audio; step up if the transcript has errors.
English-only variants are more accurate than multilingual ones at the same size when the audio is definitely English. Use a multilingual model when the language is not English or when you want to translate speech into English.
Questions
- Is my audio uploaded for transcription?
- No. The model runs in your browser and the audio stays on your device. Nothing is sent to a transcription service.
- Why is the first transcription slow?
- The model weights have to download before inference can start. That is a one-time cost per model — later runs reuse the cached copy.
- Can it transcribe a video file?
- Yes. The audio track is extracted from the video in the browser and transcribed the same way.