ReelDeal

Turns Reddit stories into short-form vertical videos with AI voiceover and captions — ready to post on YouTube Shorts, TikTok, and Instagram Reels.

Example

Source: AITA for refusing to give my mom my savings?

• YouTube Shorts
• Instagram Reels
• TikTok

How it works

1. Scrape a Reddit post and clean the text via Claude CLI
2. Generate voiceover audio (ElevenLabs or Edge TTS)
3. Compose a 9:16 video with gameplay background, word-by-word captions (Whisper), and the voiceover
4. Generate platform-specific titles, descriptions, and hashtags
5. Track post performance across platforms with the analytics dashboard

Pipeline architecture

Each story lives in output/<slug>/ with a single story.json containing all structured data (source URL, raw text, cleaned transcript, platform metadata). Binary files (audio.mp3, captions.ass, final.mp4) sit alongside it.

Batch processing writes a batch_manifest.json to track approved/filtered/failed stories across runs.

Setup

pip install -r requirements.txt

Drop gameplay footage (.mov/.mp4) into gameplay/ — the pipeline picks a random clip as background.

Environment variables

Variable	Required	Description
ELEVENLABS_API_KEY	No	ElevenLabs API key for premium TTS
ELEVENLABS_VOICE_ID	No	ElevenLabs voice ID (e.g. Brian: nPczCjzI2devNBz1zQrb)

If ElevenLabs is not configured, the pipeline falls back to Edge TTS (free, no API key needed) with automatic gender-based voice selection.

Usage

# Full pipeline: Reddit URL → finished video
python cli.py full <reddit-url>

# Or run steps separately
python cli.py text <reddit-url>     # scrape + voiceover + metadata
python cli.py video <slug>          # compose video from output/

# Batch pipeline (recommended for multiple stories)
python cli.py batch-prep urls.txt   # scrape + filter + generate manifest
python cli.py batch-clean           # parallel transcript cleaning via Claude CLI

# Register uploads for analytics tracking
python cli.py register <slug> --youtube <url> --tiktok <url> --ig <url>