Agent-Skills.md

Agent Skills: Caption Format Conversion

Use when converting between caption formats (SRT, VTT, ASS, TTML, Gemini MD, etc.). Supports 30+ caption formats.

UncategorizedID: lattifai/omni-captions-skills/omnicaptions-convert

Repository

lattifai/omni-captions-skills
lattifaiLicense: MIT
222

Install this agent skill to your local

pnpm dlx add-skill https://github.com/lattifai/omni-captions-skills/tree/HEAD/skills/omnicaptions-convert

Skill Files

Browse the full folder contents for omnicaptions-convert.

Download Skill

Loading file tree…

skills/omnicaptions-convert/SKILL.md

Skill Metadata

Name
omnicaptions-convert
Description
Use when converting between caption formats (SRT, VTT, ASS, TTML, Gemini MD, etc.). Supports 30+ caption formats.

Caption Format Conversion

Convert between 30+ caption/caption formats using lattifai-captions.

⚡ YouTube Workflow

# 1. Transcribe YouTube video directly
omnicaptions transcribe "https://youtube.com/watch?v=VIDEO_ID" -o transcript.md

# 2. Convert to any format
omnicaptions convert transcript.md -o output.srt
omnicaptions convert transcript.md -o output.ass
omnicaptions convert transcript.md -o output.vtt

When to Use

  • Converting SRT to VTT, ASS, TTML, etc.
  • Converting Gemini markdown transcript to standard caption formats
  • Converting YouTube VTT (with word-level timestamps) to other formats
  • Batch format conversion

When NOT to Use

  • Need transcription (use /omnicaptions:transcribe)
  • Need translation (use /omnicaptions:translate)

Setup

pip install omni-captions-skills --extra-index-url https://lattifai.github.io/pypi/simple/

Quick Reference

| Format | Extension | Read | Write | |--------|-----------|------|-------| | SRT | .srt | ✓ | ✓ | | VTT | .vtt | ✓ | ✓ | | ASS/SSA | .ass | ✓ | ✓ | | TTML | .ttml | ✓ | ✓ | | Gemini MD | .md | ✓ | ✓ | | JSON | .json | ✓ | ✓ | | TXT | .txt | ✓ | ✓ |

Full list: SRT, VTT, ASS, SSA, TTML, DFXP, SBV, SUB, LRC, JSON, TXT, TSV, Audacity, Audition, FCPXML, EDL, and more.

CLI Usage

# Convert (auto-output to same directory, only changes extension)
omnicaptions convert input.srt -t vtt           # → ./input.vtt
omnicaptions convert transcript.md              # → ./transcript.srt

# Specify output file or directory
omnicaptions convert input.srt -o output/       # → output/input.srt
omnicaptions convert input.srt -o output.vtt    # → output.vtt

# Specify format explicitly
omnicaptions convert input.txt -o out.srt -f txt -t srt

ASS Style Presets

When converting to ASS format, use --style to apply preset styles:

omnicaptions convert input.srt -o output.ass --style default    # White text, bottom
omnicaptions convert input.srt -o output.ass --style top        # White text, top
omnicaptions convert input.srt -o output.ass --style bilingual  # White + Yellow (for bilingual)
omnicaptions convert input.srt -o output.ass --style yellow     # Yellow text, bottom

| Preset | Position | Line 1 | Line 2 | Use Case | |--------|----------|--------|--------|----------| | default | Bottom | White | White | Standard captions | | top | Top | White | White | When bottom is occupied | | bilingual | Bottom | White | Yellow | Bilingual captions (原文 + 译文) | | yellow | Bottom | Yellow | Yellow | High visibility |

Bilingual Example

If your SRT has two-line captions like:

1
00:00:01,000 --> 00:00:03,000
Hello World
你好世界

Use --style bilingual or custom colors:

# Preset: white + yellow
omnicaptions convert bilingual.srt -o output.ass --style bilingual

# Custom colors: green English + yellow Chinese
omnicaptions convert bilingual.srt -o output.ass --line1-color "#00FF00" --line2-color "#FFFF00"

# Mix preset with custom line2 color
omnicaptions convert bilingual.srt -o output.ass --style default --line2-color "#FF6600"

Custom Color Options

| Option | Description | |--------|-------------| | --line1-color "#RRGGBB" | First line (original) color | | --line2-color "#RRGGBB" | Second line (translation) color |

Common colors: #FFFFFF (white), #FFFF00 (yellow), #00FF00 (green), #00FFFF (cyan), #FF6600 (orange)

Font Size and Resolution

Font size is auto-calculated based on video resolution. Resolution is detected from (priority order):

  1. --resolution argument (e.g., 1080p, 4k, 1920x1080)
  2. --video argument (uses ffprobe to detect)
  3. .meta.json file (saved by omnicaptions download)
  4. Default: 1080p
# Auto-detect from .meta.json (saved by download command)
omnicaptions convert abc123.en.srt -o abc123.en.ass --karaoke

# Specify resolution directly
omnicaptions convert input.srt -o output.ass --resolution 4k
omnicaptions convert input.srt -o output.ass --resolution 720p
omnicaptions convert input.srt -o output.ass --resolution 1920x1080

# Detect from video file (uses ffprobe)
omnicaptions convert input.srt -o output.ass --video video.mp4

# Override auto-calculated fontsize
omnicaptions convert input.srt -o output.ass --resolution 4k --fontsize 80

| Resolution | PlayRes | Auto FontSize | |------------|---------|---------------| | 480p | 854×480 | 24 | | 720p | 1280×720 | 32 | | 1080p | 1920×1080 | 48 (default) | | 2K | 2560×1440 | 64 | | 4K | 3840×2160 | 96 |

Karaoke Mode

Generate karaoke subtitles with word-level highlighting. Requires word-level timing (use LaiCut alignment first).

# Basic karaoke (sweep effect - gradual fill)
omnicaptions convert lyrics_LaiCut.json -o lyrics_LaiCut_karaoke.ass --karaoke

# Different effects
omnicaptions convert lyrics_LaiCut.json -o lyrics_LaiCut_karaoke.ass --karaoke sweep    # Gradual fill (default)
omnicaptions convert lyrics_LaiCut.json -o lyrics_LaiCut_karaoke.ass --karaoke instant  # Instant highlight
omnicaptions convert lyrics_LaiCut.json -o lyrics_LaiCut_karaoke.ass --karaoke outline  # Outline then fill

# LRC karaoke (enhanced word timestamps)
omnicaptions convert lyrics_LaiCut.json -o lyrics_LaiCut_karaoke.lrc --karaoke

| Effect | ASS Tag | Description | |--------|---------|-------------| | sweep | \kf | Gradual fill from left to right (default) | | instant | \k | Instant word highlight | | outline | \ko | Outline fills, then text fills |

Karaoke Workflow

# 1. Align with LaiCut (get word-level timing in JSON)
omnicaptions LaiCut audio.mp3 lyrics.txt

# 2. Convert to karaoke ASS
omnicaptions convert lyrics_LaiCut.json -o karaoke.ass --karaoke

# Or combine with style
omnicaptions convert lyrics_LaiCut.json -o karaoke.ass --karaoke --style yellow

Python Usage

from omnicaptions import Caption

# Load any format
cap = Caption.read("input.srt")

# Write to any format
cap.write("output.vtt")
cap.write("output.ass")
cap.write("output.ttml")

Common Mistakes

| Mistake | Fix | |---------|-----| | Format not detected | Use --from / --to flags | | Missing timestamps | Source format must have timing info | | Encoding error | Specify encoding="utf-8" |

Related Skills

| Skill | Use When | |-------|----------| | /omnicaptions:transcribe | Need transcript from audio/video | | /omnicaptions:translate | Translate with Gemini API | | /omnicaptions:translate | Translate with Claude (no API key) | | /omnicaptions:download | Download video/captions first |

Workflow Examples

# Transcribe → Convert → Translate (with Claude)
/omnicaptions:transcribe video.mp4
/omnicaptions:convert video_GeminiUnd.md -o video.srt
/omnicaptions:translate video.srt -l zh --bilingual