Caption Styles That Work for Podcast Clips (with API Examples)

Most podcast clips live or die on the first two seconds, and the captions carry them. The style you pick decides whether someone reads along or keeps scrolling. This guide walks through the caption styles that consistently work for talking-head podcast clips and shows the exact caption.sh API options for each one.

Start with word count, not font

The biggest lever is how many words are on screen at once. One to three words per group reads like karaoke and keeps eyes moving. Five or more words per group reads like a subtitle and suits slower, interview-heavy clips.

In caption.sh this is controlled by max_words and max_lines. max_words caps how many words appear in each caption group, and max_lines caps how many lines that group can wrap to before it splits.

{
  "source_url": "https://example.com/episode-42.mp4",
  "max_words": 3,
  "max_lines": 1
}

The three styles that work

After captioning a lot of podcast clips, three styles cover almost every show.

• Bold karaoke: one to three words at a time, large size, center position, outline decoration. Best for short punchy moments and clips under 30 seconds.
• Lower third subtitle: four to six words, small or medium size, bottom position, backdrop decoration. Best for longer clips where the guest's face matters more than the words.
• Top bar hook: two to four words, medium size, top position. Keeps the caption out of the platform's UI zone and pairs well with a strong first line.

Position and the safe zone

TikTok, Reels, and Shorts all paint UI over the right edge and the bottom of the frame. A bottom caption on TikTok sits under the description text. For clips headed to every platform, center or top positions survive cropping better. caption.sh takes position as top, center, bottom, or a fraction of the video height, so position 0.78 puts the caption just above the description zone.

{
  "source_url": "https://example.com/episode-42.mp4",
  "font": "luckiest-guy",
  "size": "large",
  "position": 0.78,
  "max_words": 4,
  "decoration": "backdrop"
}

Font and decoration choices

caption.sh bundles 18 fonts. For podcasts, the heavy display and slab faces win because they stay readable over busy backgrounds and small screens: luckiest-guy (the default) for the bold karaoke look, ultra or arvo for a slab feel, fredoka for a softer rounded sans. Decoration is the layer behind the text: outline draws a stroke around each glyph, backdrop draws a solid block behind the words, and none leaves the text bare. Outline reads best over talking heads. Backdrop reads best over b-roll and screen recordings.

Size accepts small, medium, large, or a fraction of the video height. Fractions scale with the source, so size 0.06 looks the same on a 720p and a 4K render.

A complete example

Here is a full request for a bold karaoke style clip. The API transcribes the audio, times every word, and returns a job you poll until the rendered video is ready.

curl -X POST https://api.caption.sh/v1/videos \
  -H "Authorization: Bearer $CAPTION_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "source_url": "https://example.com/episode-42.mp4",
    "font": "luckiest-guy",
    "size": "large",
    "position": "center",
    "max_words": 3,
    "max_lines": 1,
    "decoration": "outline"
  }'

Pricing is per minute of video at $0.10 per minute, so a 90 second podcast clip costs $0.15 to caption. Start with the bold karaoke defaults, then adjust max_words and position until the clip reads the way your show sounds.