ENGLISH AUDIO / VIDEO → SUBTITLES
Auto captions.
Choose audio or video
One English WAV, MP3, MP4 or WebM file. Tested with 3- and 10-minute audio; browser codec and memory limits vary.
Choose a file to begin.
Your media never leaves your browser. Only model and runtime files are downloaded; no audio or video bytes are uploaded.
Timed captions
Windows that could not be recognized
Intervals with no recognized caption
These intervals may be silence or speech the model missed. Listen to the source and fill any missing words yourself.
Before you start
This version recognizes English only. It processes one complete file in overlapping 30-second windows and joins timestamps into one SRT or VTT. There is no fixed duration or file-size cutoff in this tool; your browser's codec, memory and available storage can still limit a job. A video plays locally at normal speed while its audio is captured, before recognition begins. A file with no audible audio cannot produce captions.
On our test Windows machine, a cached-model run took 51 seconds for 3 minutes of continuous English speech and 168 seconds for 10 minutes. Those samples were synthetic English speech; your device, media and first model download will differ. The 8-bit Whisper Tiny English encoder and decoder weights total 40,843,851 bytes (38.95 MiB), measured from the pinned model files, plus tokenizer and runtime files. The model downloads from Hugging Face on first use and is stored in browser cache when space allows.
Timing comes from the model and may be imprecise. Captions are editable above before download. Intervals without recognized captions are listed separately; they are not automatically filled. Review names, punctuation, long cues and quiet speech against the original media. For paragraph text without timing, use video to text.
To edit a subtitle file you already have, use the subtitle editor. For other tasks, use VTT to SRT, SRT to VTT, Text to SRT, SRT to TXT, VTT to TXT, the subtitle converter, ASS to SRT, SUP to SRT, or the CPS checker.