How to Add Subtitles and Captions to AI-Generated Videos

Around 80% of short-form video watchers prefer to view their videos on their mobile devices with the sound turned off. Therefore, using appropriate, synchronized subtitles in your AI-generated video is key in order to keep the attention of viewers, increase your content's search rankings, and make your videos inclusive.

This section gives you a practical guide for generating, styling, and synchronizing the subtitles to your AI videos across all major platforms.

1. The 5-Step Subtitle & Captioning Workflow

To auto-generate your subtitles through automated captioning for your videos, follow these five steps:

1. Export Audio Files from Text-to-Speech Technology

• Create your audio files in one of the text-to-speech applications available on the market today and keep audio files free of long spaces during speeches.

2. Run Auto Speech Recognition (ASR)

• Import your master video and audio into your caption tool Cap CapCaput, Descript, or Whisper. Run the auto-transcription process so that the tool converts the speech to time-coded segments.

3. Audit Spelling, Proper Names, & Abbreviations

• Examine the transcribed content carefully line by line and make necessary amendments regarding specific domain terms, product names, and errors made by phonetic speech recognition (for example, “SaaS” should not be written as “sauce”).

4. Apply Safe Zone Margin Alignment

• Put all caption text in Middle-Third Safe Zone (35% to 65% of frame in the Y-direction). Line breaks must be short (max. 2 lines per screen, max. 32 to 40 characters per line).

5. Apply Dynamic Contrast & Render Master

• Add a border around text so it stands out against footage (strong outline of white text with 2px thick black border or dark shadow box). Create a MP4 with hardcoded caption text or store an. Srt sidecar.

2. Avoiding Mobile UI Trapping: Safe Margins Guide

Placing captions incorrectly on vertical (9:16) videos causes them to be hidden behind native platform interfaces:

+---------------------------------------+
|  [TOP SAFE ZONE - 15% PADDING]        | ➔ Avoid: Obscured by Search Bar / Header
+---------------------------------------+
|                                       |
|  ✅ RECOMMENDED CAPTION ZONE           | ➔ Safe Viewport: 35% to 65% Screen Height
|     (Word-by-word highlights)         |
|                                       |
+---------------------------------------+
|  [RIGHT: Likes/Share Icons]           | ➔ Avoid: Obscured by Right UI Action Rails
|  [BOTTOM: Username / Sound Info]      | ➔ Avoid: Obscured by Bottom Description & Seekbar
+---------------------------------------+

• Vertical Top Padding: Ensure that your vertical top padding space is 150px (15%) from the top.
• Vertical Bottom Padding: Ensure that your vertical top padding space is 250px (25%) from the bottom edge to avoid overlapping with handles of platforms captions and audioTitles.
• Horizontal Margin: Ensure to keep 10% side padding to avoid any overlap with action buttons on the right-side such as like, comments, and bookmarks.

3. Advanced SubStation Alpha (.ass) Styling Engine

When comparing .srt files, which enable only the most basic time stamps and standard text, the Advanced SubStation Alpha (.ass / .ssa) format gives you the possibility to create your own typography, positional vectors, karaoke-like highlighting, and transformations without even having to use a desktop video editing application.

• [Raw Audio file] ➔ [Faster-Whisper in audio file (words time stamps)] ➔ [Python .ASS Formatter] ➔ [FFmpeg libass Render Process] ➔ [Final Video in MP4 format]

Breakdown of Key ASS Header Parameters:

• Alignment: 2: Centers text and is anchored to the vertical Baseline dictated by MarginV.
• MarginV: 850: At 1080x1920, this creates a 850px margin at the bottom of the screen, ensuring text rests in the sweet spot middle of the screen (y = ~1070px from top) so it will be out of the path of native social overlays.
• Outline: 8 & Shadow: 12: Creates an 8px solid black border and a 12px drop shadow to ensure it’s easily readable against any background image.
• {\k(duration)}: This is how long in centiseconds (hundredths of a second) the karaoke color swing should persist for that word.

4. Headless Video Burn-In via FFmpeg (libass)r5t

Burning .ass kinetic subtitles directly into the master video to avoid re-encoding and loss of quality, with a FFmpeg compiled with libass:

ffmpeg -y -i raw_input_video.mp4 \
  -vf "ass=captions_master.ass" \
   -c:v libx264 -preset slow -crf 18 \
  -c:a copy \
  final_master_burned_video.mp4

• -vf "ass=captions_master.ass": Uses the hardware-accelerated libass rasterizer to draw text, vector strokes, and karaoke wipes directly onto video frames.
• -crf 18: Constant Rate Factor set to 18 ensures visual lossless mastering for high-definition social distribution.
• -c:a copy: Stream-copies the audio track directly without re-compression to prevent acoustic degradation.

5. Subtitling Accessibility and Readability Standards (WCAG 2.2)

If you use subtitles for media hubs, web players, or video ads, you need to adhere to these Web Content Accessibility Guidelines (WCAG 2.2 Level AA) standards:

• Contrast Ratio (minimum 4.5:1): Text overlays of captions against dynamic video playback must have a minimum contrast ratio of 4.5:1 (via an opaque bounding box or a black outline of 8px).
• Reading Speed Limitations: Limit the density of the text to 17-20 characters per second (CPS) since captions that are moving at a speed of 22 CPS and more will cause too much cognitive overload for the audience.
• Limit on the number of characters per line: The maximum number of characters in the line should be no more than 37 on vertical displays in a ratio of 9:16 and no more than 42 characters on horizontal monitors in a ratio of 16:9.