English sample · 10.56 s
Dataset reference text
However, due to the slow communication channels, styles in the west could lag behind by 25 to 30 year.
The reference is a dataset transcript, not a promise of the tool's output.Large-file transcription
Transcribe a large audio or video file on your device
Choose a local file up to 1 GB and 1 hour. Whisper Web decodes and transcribes it in smaller sections instead of loading the entire recording into memory.
Local file only · up to 1 GB and 1 hour · no media uploadProgress is kept only in this tab. Reloading, closing the tab or putting the device to sleep means starting again.
From a long recording to an editable transcript
Choose the original file
Use a local audio or video file up to 1 GB and 1 hour. A browser check confirms its audio track and duration before you start.
Choose the spoken language
Start with Tiny and WebAssembly for a smaller download. Larger models need more memory and may take longer.
Keep this tab open
The tool processes five-minute sections. Pause finishes the current section; Resume continues with the next one. Closing the tab loses unfinished progress.
Review and export
Check names, numbers and uncertain passages against the recording. Export TXT, SRT or VTT; save a copy before leaving the page.
Check a short recording before a long job
Format support depends on the audio codec inside the file and your browser. Try a current desktop browser for a long recording. If decoding fails, export the audio as MP3, WAV or M4A and check it again. WebGPU is optional; WebAssembly remains available as the default.
Related guides
Try a short recording before your own file
Listen to a public English or Spanish sample, load it into the tool, then compare your result with the dataset reference. These short read-speech clips do not represent noisy meetings or long recordings.
Spanish sample · 12.84 s
Dataset reference text
Se recomienda enfáticamente a los viajeros que se informen sobre cualquier riesgo de clima extremo en el área que visitan, dado que ello puede afectar sus planes de viaje.
The reference is a dataset transcript, not a promise of the tool's output.Audio and reference: FLEURS contributors / Google, CC BY 4.0. Original recordings are unchanged. FLEURS · CC BY 4.0
Sample files and browser test resultsLong local jobs can warm up your device
Whisper uses the CPU or GPU for an extended period. Higher power use, fan noise and a warmer device can occur during processing.
- Connect the device to power for a long recording.
- Keep ventilation clear and stop if the device becomes unusually hot.
- Leave this tab open and keep the device awake until the transcript is saved.
Questions about large-file transcription
Does the large file get uploaded?
No. The browser reads, decodes and transcribes the selected file on this device. It still downloads the site and the selected Whisper model.
Why can a large transcription take so long?
The work runs on your CPU or GPU and speed varies by device, browser, model and power mode. The page calculates a reference range after the first processing block.
Can I close the tab and return later?
No. Partial progress is kept only in this tab. Keep it open and keep the device awake until the completed transcript is saved.
What happens when I pause?
The current five-minute section finishes first. Processing then stops between sections, and Resume continues with the next one.
Where is the completed transcript stored?
It is saved to IndexedDB in this browser, where you can edit it and export TXT, JSON, SRT or VTT.