VideoMaker – Automated TikTok Video Generator

VideoMaker is a fully automated, extensible system for generating, editing, and publishing short-form videos. It supports multiple content sources (Reddit, YouTube, text splits), text-to-speech engines, video composition, audio mixing, and automated TikTok posting via Selenium.

The system is built in Python and orchestrates a full pipeline from content ingestion to final publishing.

---

Features

Content Builders

• Reddit Stories Builder

  • Fetches posts/comments using PRAW.
  • Translates (using deep_translator) and sanitizes text.
  • Generates narration via multiple TTS providers.
  • Creates vertical videos using YouTube clips as background.

• Simple Splits Builder

  • Downloads videos from YouTube playlists.
  • Detects highlight markers with the LemnosLife API.
  • Extracts interesting segments or fallback random chunks.
  • Creates split-screen videos with titles overlaid.

Media Pipeline

• Audio generation

  • Google Cloud TTS
  • TTSMP3
  • gTTS
  • Auto speed-adjustment, gap insertion, and merging.

• Video generation

  • Background extraction and cropping (9:16 format).
  • Layering text, audio, and overlay clips.
  • Automatic subtitle line breaking.
  • Final rendering via MoviePy (H.265 codec).

Automation & Persistence

• Reddit sampling history tracking.
• YouTube video usage tracking.
• Safe persistence of produced videos to avoid duplicates.
• Per-account generation settings.

TikTok Auto-Posting

• Headless Chrome automation with cookie-based login.
• Multi-account support.
• Robust handling of upload screens and rendering delays.

---

Project Structure

mastertuto-videomaker/
│
├── main.py                        # Orchestrates full run: generation + posting
├── requirements.txt
│
├── content/
│   ├── builders/                  # RedditStoriesBuilder, SimpleSplitsBuilder
│   ├── configs/                   # JSON profiles per channel/subreddit
│   ├── reddit/                    # Reddit fetching logic
│   ├── youtube/                   # YouTube playlist/handler tools
│   └── markers/                   # Detection of YouTube "most replayed" markers
│
├── media/
│   ├── audio/                     # TTS + audio mixing
│   ├── video/                     # VideoMaker (composition, overlays, rendering)
│   ├── text/                      # Text processing & layout
│   ├── stt/                       # Future speech-to-text subsystem
│   └── tts/                       # TTS providers (Google, gTTS, TTSMP3)
│
├── tiktok/                        # Automated posting
├── translate/                     # High-level translation interface
├── exceptions/
└── utilities/                     # Config, logging, persistence, helper functions

---

Installation

Requirements

• Python 3.10+
• Chrome installed
• A Google Cloud project (optional but recommended for TTS)
• YouTube API not required (Pytube only)
• Reddit API credentials for PRAW

Install dependencies

pip install -r requirements.txt

---

Configuration

1. Reddit API Credentials

Edit utilities/config/reddit.py:

CLIENT_ID = "your_id"
CLIENT_SECRET = "your_secret"

2. TikTok Cookies

Generate and store cookies using browser extensions like Cookie-Editor, then place each account’s cookie file in your config folder.

3. Video Config Profiles

You can create infinite accounts in:

utilities/config/accounts.py

Each entry defines:

• Content type (reddit_stories or simple_splits)
• Source config file
• Voice language
• Cookie file path

Example:

{
    "type": "reddit_stories",
    "lang": "pt-br",
    "login_file": "cookies/account1.json",
    "videos_config": ["askreddit_pt.json", "nosleep_pt.json"]
}

---

Usage

Simply run:

python main.py

The system will:

1. Process pending videos from previous runs.

2. Shuffle accounts.

3. For each account:

   • Select two random video configs.
   • Build videos.
   • Save them.
   • Attempt to auto-post to TikTok.

All outputs are written under the configured OUTPUT_FOLDER.

---

Adding New Content Types

You can extend the platform by implementing new builders:

content/builders/

Each builder must expose:

• build(persistence, translate_to)
• deploy()

---

Developing CSS or Visual Layers

If you intend to add styling overlays, dynamic captions, or CSS-like layout rules, you may consider extending:

media/text/editor.py
media/video/maker.py

The repository is structured so that visual formatting logic is isolated from content sourcing.

---

Roadmap

• Automatic caption generation (STT)
• Background music synchronization by BPM
• Template-based rendering
• OpenAI/GPT support for story rewriting
• Full web dashboard