Speech-to-text tool
Create timed subtitles from speech in your video. Everything happens in your browser.
Ready on your device
Choose a video you have the right to use. It is not uploaded.
Start here
Create timed subtitles from spoken English, Spanish, French, or Chinese.
Drop a video here
or
Use a video you own or have permission to process. Clear speech gives the best result.
Selected video
Choose the language spoken in the video. This tool transcribes speech; it does not translate it.
Whisper Base runs locally
First use downloads about 77 MB of model data from the NoUpload App model host. Your selected video is not sent with those requests.
The speech model is coming to your browser. Your video remains on your device.
Finished on your device
Review the transcript, then save the format you need.
Audio decoding and Whisper transcription happen in this browser. Your video is not uploaded or stored on a processing server.
A clear local workflow
Choose a permitted video, identify the spoken language, and let Whisper work on your device.
Select a video that you own or have permission to process. Choosing the file does not upload it. A preview and basic details appear after the browser reads the file. Clear speech with limited background noise generally produces the most useful transcript.
Select English, Spanish, French, or Chinese. Whisper uses that choice to transcribe the original speech rather than translate it. The browser decodes the audio track and resamples it for the speech model without sending the video away.
Review the timed text because names, numbers, accents, and noisy speech can be misheard. Save SRT for common editors and players, VTT for web video, or plain TXT when you only need the transcript.
Whisper Base can handle ordinary speech, but no automatic transcript is perfect. Quiet recordings, one speaker at a time, and clear microphones help. Music, overlapping voices, strong accents, technical names, low volume, and compressed audio can reduce accuracy. Always review subtitles before publishing them, especially when exact names, dates, prices, medical terms, or legal wording matter.
The Whisper model comes to your browser from NoUpload App's dedicated model host; your chosen video does not go to a transcription server. The first use downloads roughly 77 MB of ONNX model data plus the local runtime files. Your browser may cache these assets for faster reuse. Long videos need substantial memory and time, so begin with a short video while checking browser compatibility.
Straight answers
Useful details before you transcribe a video.
No. The browser reads your video and decodes its audio locally. Whisper then creates the transcript in this tab. Selecting the file only gives this browser tab permission to read it; the video is not uploaded to us.
Speech recognition needs the quantized Whisper Base encoder and decoder. Those model files download to your browser before transcription. The transfer direction is toward your device; your selected video is not sent with those requests.
The launch interface supports spoken English, Spanish, French, and Chinese. Choose the language actually spoken in the video. The tool transcribes that language and does not translate it into another language.
You can save SubRip SRT, WebVTT VTT, or plain TXT. SRT works with many players and editors. VTT is designed for web video. TXT contains the spoken text without subtitle timing syntax.
No. Use only files you own or have permission to process. This tool does not provide a downloader and is not intended to bypass copyright, platform rules, access controls, or another person's privacy.
No account or sign-in is required, and the tool is free to use.
More local tools
