Generate captions from the audio

Import your video into ClippingShorts Studio, then choose Generate captions. Select the current clip or the full video and choose the spoken language, or let the model detect it. Start with a short clip to check accuracy and processing speed on your device.

Speech recognition runs locally after the model downloads. Your audio is not uploaded for transcription. Keep the tab open while the browser works. A smaller multilingual model can miss names, accents, overlapping speech or quiet audio, so generation should always be followed by review.

Import SRT or VTT subtitles when you already have them

If you have a corrected transcript from another tool, use Import SRT / VTT. These files include timestamps that place each caption on the source timeline. Import subtitles timed to the original recording, not an unrelated edited version.

Play several points near the beginning, middle and end to check alignment. A subtitle file can contain the right words but still drift if the source has a different start time, missing sections or a different duration. Do not assume a successful import means the timing is correct.

Correct the words before choosing a style

Open the Text tab and click a timestamp to seek to that caption. Listen to the sentence and edit its wording directly. Prioritize names, dates, quantities, technical vocabulary and words that affect the meaning of the statement.

Keep punctuation natural and avoid unnecessary all-caps text. Check Urdu, Hindi and other multilingual passages with particular care: code-switching and proper nouns can be difficult for a small speech model. Imported subtitles or manual corrections can provide a better result when automatic recognition struggles.

Make captions readable in a small video

Choose from bold, minimal, boxed or karaoke-style captions. Adjust the text size, vertical position and highlight colour. A high-contrast treatment helps text remain readable as the footage behind it changes. Check a bright frame and a dark frame rather than styling against just one still.

Keep captions away from important faces, product details and screen edges. Watch the clip at a small preview size to judge whether the words remain readable. Word highlighting from segment-only subtitles uses estimated timing, so it should not be treated as precise word alignment.

Choose burned-in captions or a subtitle file

When you export a video, the styled captions are drawn into its frames. This keeps their appearance with the video, but viewers cannot switch them off. You can also download an SRT subtitle file for the selected clip, with times shifted to the clip’s beginning.

Play the final export to check caption visibility and audio sync. Save your project JSON alongside the original video if you want to make changes later. For the complete editing sequence, follow the long video to shorts guide.

Put it into practice

Keep learning

How to turn a long video into short clips ↗How to make podcast clips that stand on their own ↗