# How do I add captions to Shorts automatically?

Canonical page: https://remixviral.ai/en/solutions/auto-captions-for-shorts | Last updated: 2026-11-13 | By RemixViral (https://remixviral.ai)

Generate a transcript with speech recognition, correct names and numbers by hand, then either let the platform show its own automatic captions or burn the text into the video so it always shows. YouTube generates automatic captions for Shorts but says to review them. A clip maker or ffmpeg can burn captions in. Keep text away from the bottom edge.

## Why this happens

Short videos are often watched without sound, and viewers who are deaf or hard of hearing need text regardless. That is why captions are a standard part of the Shorts workflow, and why people want them added automatically rather than typed line by line.

"Captions" means two different things, and the method you pick depends on which you need. Platform captions are a separate text track the platform displays and the viewer can switch off. Burned-in captions (sometimes called open captions) are drawn into the picture itself, so they appear on every platform and in every player, and they can carry a style such as a highlighted current word. The price of burning in is that the text is now pixels: a typo cannot be fixed without re-rendering.

Automatic speech recognition is good but not perfect. YouTube's own help page on automatic captioning (read 2026-11-13) says the captions are generated by machine learning, that quality may vary, and that they might misrepresent the spoken content because of mispronunciations, accents, dialects or background noise. It tells creators to always review automatic captions and edit any parts that were not properly transcribed. The same advice applies to any tool, including ours.

## How to add captions step by step (works without any particular tool)

Plan on a few minutes per clip for review. The review is the step that decides whether the captions look professional.

1. **Decide: platform captions, burned-in captions, or both.** Platform captions are the right default for accessibility and for a video you upload to one platform, because viewers can turn them on or off and the text can be corrected after publishing. Burned-in captions are the right choice when the clip will be posted on several platforms, when you want a style (large words, a highlighted current word), or when you cannot rely on the platform to show captions. Using both can show the text twice; if you burn in, you may prefer to leave the platform track off or to keep it only for accessibility and search.

2. **Get a transcript.** For YouTube, upload the Short and let YouTube generate automatic captions. Per its help page, automatic captions are published on the video when available, may not be ready right after upload, and depend on the audio: it lists reasons they may not appear, including an unsupported language, a long silence at the start, poor sound quality, or overlapping speakers. Any speech-to-text tool or an AI clipper can also produce a transcript with timestamps.

3. **Review names, numbers and brand terms.** Read the captions against the audio, not on their own. Errors cluster in proper nouns, product names, numbers, units and homophones. In YouTube Studio, open Subtitles, pick the video, and edit the automatic captions there. If you plan to burn in, fix the text before rendering: a wrong number baked into the picture stays wrong.

4. **Export or write the captions as an SRT file.** SRT is plain text. Each caption is a counter line, a time range in HH:MM:SS,mmm --> HH:MM:SS,mmm with a comma before the milliseconds, one to two lines of text, and a blank line. Times are relative to the start of the clip, so if your clip starts 5 minutes into the source, the first caption you want should be near 00:00:01, not 00:05:01.

   ```
   1
   00:00:01,000 --> 00:00:03,500
   The first rule of cold email
   
   2
   00:00:03,500 --> 00:00:06,000
   is that nobody reads long ones.
   ```

5. **Burn the captions in with ffmpeg.** This command, from our tested ffmpeg post, cuts 10 seconds starting at the 5-second mark, centre-crops to vertical 1080x1920, draws captions.srt with the libass-based subtitles filter, and encodes for upload. It needs an ffmpeg build with libass and libx264. Alignment=2 is bottom-centre, MarginV=120 lifts the text off the bottom edge and Outline=2 adds a border; tune FontSize by eye on a test frame.

   ```
   ffmpeg -ss 5 -i input.mp4 -t 10 \
     -vf "crop=ih*9/16:ih:(iw-ih*9/16)/2:0,scale=1080:1920,setsar=1,subtitles=captions.srt:force_style='FontName=Arial,FontSize=14,Alignment=2,MarginV=120,Outline=2'" \
     -c:v libx264 -preset medium -crf 20 -pix_fmt yuv420p \
     -c:a aac -b:a 128k -movflags +faststart clip.mp4
   ```

6. **Check placement and watch the whole clip.** Play it on a phone in the app you are posting to. Practical guidance, from habit rather than from any platform document we have read: keep captions in the middle band of the frame and off the bottom edge, where the app usually draws the title, account name and buttons, and keep the text inside the side margins. Two lines at most, and enough contrast against the background (an outline or a dark box helps on busy footage).

## Captions and accessibility

The W3C's explanation of WCAG Success Criterion 1.2.2, Captions (Prerecorded) (Level A), read on 2026-11-13, says captions should be provided for all prerecorded audio content in synchronized media, and that the intent is to let people who are deaf or hard of hearing watch it. It also says captions include dialogue and identify who is speaking, and include non-speech information conveyed through sound, including meaningful sound effects.

That has a practical consequence: a transcript of only the words may fall short of the criterion on a video where sound effects or speaker changes carry meaning. Burned-in captions are visible to everyone but cannot be turned off or restyled by the viewer, so for accessibility-focused publishing a separate caption track alongside is worth considering. We are not accessibility auditors; check the standard that applies to you.

## A short review checklist

Before you export or upload, check each clip once:

- Names of people, companies and products spelled correctly.
- Every number, date and unit matches what is said.
- Line breaks fall at natural phrase boundaries, two lines at most.
- Captions are clear of the bottom of the frame and the screen edges.
- Text contrast is readable over the busiest part of the clip.
- The first caption appears when the speaker starts, not a beat late.

## When to use RemixViral

RemixViral burns captions into every clip it cuts from a video you upload, with no separate caption step. It transcribes the whole video, picks and ranks moments, cuts each to vertical 9:16 (or 1:1 or 16:9) and renders the captions into the picture. You choose a caption style: classic (yellow highlight), plain white, neon (cyan highlight), green highlight, compact, or top of frame. Source language is auto-detected; we do not publish a list of supported languages, so test a short sample in your language first.

Use it when:

- You want captions in every clip from a long video you own, in one of six styles, without a manual transcript step.
- You want to review each clip, in the dashboard, for caption errors before you export; names and numbers are the usual misses.
- You post the same clip on several platforms and want the captions to travel with the video.

It is not the right tool when:

- You need captions you can edit word by word after export: RemixViral renders them into the picture, and it has no timeline or frame-accurate editor. Use a caption editor that exports a separate caption file, or fix the text in your own editor.
- You need a standards-grade accessibility caption track (speaker labels, sound-effect labels): use a captioning workflow or service built for that, and publish a caption file alongside the video.
- Your language may not be recognised: test first, since there is no published list.

## Ask your AI assistant to do it

After a clip job has finished, an AI assistant can read the transcript and flag likely caption mistakes for you to check in the dashboard.

```text
Assign to RemixViral: use list_clip_jobs to find my latest finished clip job, then get_clip_job with includeTranscript to read the transcript of the top three ranked clips. List every proper noun, number and brand term in those three clips so I can check how each is spelled in the burned-in captions. Do not create anything. Show me what you did so I can approve before I export.
```

Connection steps and the full tool list: https://remixviral.ai/en/mcp

## Frequently asked questions

### Does YouTube add captions to Shorts automatically?

YouTube's automatic captioning page says automatic captions are available on long-form videos and Shorts in a long list of languages and are published automatically when available. They may not be ready at upload, and YouTube says to review them because they can misrepresent speech.

### What is the difference between burned-in and automatic captions?

Automatic or platform captions are a separate text track the viewer can switch off and you can edit later. Burned-in captions are part of the picture, show everywhere and can be styled, but any typo requires re-rendering the video.

### How accurate are AI captions?

Good on clear speech and weaker on names, numbers, accents and noise; YouTube's own help page says automatic captions may misrepresent the spoken content. Always read them against the audio before publishing. We do not publish an accuracy figure for RemixViral.

### Where should captions sit on a vertical video?

Away from the bottom edge, where apps usually draw titles and buttons, and away from the sides. This is practical guidance rather than a platform specification we have read, so check a test clip in the app you post to.

## Sources

- [YouTube Help: Use automatic captioning](https://support.google.com/youtube/answer/6373554) (read 2026-11-13)
- [FFmpeg Filters Documentation (subtitles filter)](https://ffmpeg.org/ffmpeg-filters.html) (read 2026-11-13)
- [W3C WAI: Understanding Success Criterion 1.2.2 Captions (Prerecorded)](https://www.w3.org/WAI/WCAG22/Understanding/captions-prerecorded.html) (read 2026-11-13)

## Related

- [How do I turn a long video into YouTube Shorts with AI?](https://remixviral.ai/en/solutions/long-video-to-youtube-shorts)
- [ffmpeg commands: 16:9 video to vertical 9:16 Shorts](https://remixviral.ai/en/blog/ffmpeg-commands-horizontal-to-vertical-shorts)
- [How to turn a podcast into YouTube Shorts](https://remixviral.ai/en/blog/how-to-cut-a-podcast-into-youtube-shorts)
- [YouTube Shorts maker](https://remixviral.ai/en/youtube-shorts-maker)

All problems and solutions: https://remixviral.ai/en/solutions | machine-readable list: https://remixviral.ai/solutions.json