Is my video or audio uploaded?

No. The file is decoded and transcribed inside your browser. Only the open-source Whisper speech model is downloaded from Hugging Face the first time, then cached on your device.

Which languages are supported?

Whisper recognizes more than 90 languages, including English, Korean, Japanese, German, French, Spanish and Chinese. It can also translate speech from any of them into English subtitles.

How long does it take?

It depends on your device. With a GPU (WebGPU) a 10-minute file usually takes a minute or two; on CPU only it can take several times the audio length. The Fast model is quicker, the Accurate model makes fewer mistakes.

What is the difference between SRT and VTT?

Both are timed subtitle files. SRT is the most widely supported format (VLC, Premiere, YouTube). VTT is the web standard used by HTML5 video and most online players.

Extract Subtitles from MP4 & MP3 – Free AI Transcription (SRT, VTT)

How to generate subtitles from a video

Drop an audio or video file (MP3, MP4, M4A, WAV, WebM) into the box.
Choose the spoken language (it starts as this page's language). Choose “Accurate” for difficult audio.
Click Start. The first run downloads the speech model once; later runs start right away.
Check the text, then download SRT for video players, VTT for the web, or TXT for a transcript.

Subtitle formats compared

Format	Best for	Timing line example
SRT	Video players and editors (VLC, Premiere, YouTube upload)	00:00:01,000 --> 00:00:03,500
VTT	Web video (HTML5 <track>), online players	00:00:01.000 --> 00:00:03.500
TXT	Transcripts, notes, blog posts	—

Frequently asked questions

Q. Is my video or audio uploaded?: No. The file is decoded and transcribed inside your browser. Only the open-source Whisper speech model is downloaded from Hugging Face the first time, then cached on your device.
Q. Which languages are supported?: Whisper recognizes more than 90 languages, including English, Korean, Japanese, German, French, Spanish and Chinese. It can also translate speech from any of them into English subtitles.
Q. How long does it take?: It depends on your device. With a GPU (WebGPU) a 10-minute file usually takes a minute or two; on CPU only it can take several times the audio length. The Fast model is quicker, the Accurate model makes fewer mistakes.
Q. What is the difference between SRT and VTT?: Both are timed subtitle files. SRT is the most widely supported format (VLC, Premiere, YouTube). VTT is the web standard used by HTML5 video and most online players.

Related tools

Last updated: November 14, 2026