Video to text

Choose a video or audio file and press Transcribe. A speech recognition model (OpenAI's Whisper, base size) runs in your browser and writes down what's said, with timings. Download the result as plain text, SRT or VTT subtitles. The model downloads once, about 200 MB, and your file is never uploaded.

How long it takes

On a typical laptop, transcribing takes about as long as the video plays, so a 10-minute video takes around 10 minutes. A computer with a strong graphics chip is faster, because the browser runs the model on it (WebGPU). The first run also downloads the model. Keep the tab open until it finishes.

Accuracy

Whisper base is a small model. Clear speech with one speaker comes out well; heavy accents, background music and people talking over each other cause mistakes. Read the result through before you publish it. For long recordings or noisy audio, a larger model does better: ClipFlame's caption generator runs a more accurate speech model on our servers and turns the transcript into styled captions on vertical clips.

Using the subtitles

• YouTube: in Studio, open the video, then Subtitles, Add language, and upload the SRT.
• Facebook: upload the SRT under the video's Subtitles and captions settings.
• LinkedIn and web players: use the VTT.

To fix timing or convert between formats afterwards, use the subtitle converter. For a YouTube video that already has captions, pasting YouTube's own transcript into the YouTube transcript generator is quicker.

Supported files

MP4, MOV, WebM, MP3, M4A and WAV all work in Chrome, Edge and Firefox. Safari can't read the audio in some WebM files.

Common questions

Is this video to text converter free?

Yes. The speech model runs in your browser, so it's free with no file upload and no sign-up.

How accurate is it?

Clear speech from one speaker comes out well. Accents, music and people talking over each other cause mistakes, so read the result before publishing.

How long does transcription take?

About as long as the video plays on a typical laptop; faster with a strong graphics chip. The first run also downloads the model, about 200 MB.

Can I get subtitles for YouTube?

Yes. Download the SRT and upload it in YouTube Studio under Subtitles.