AI-Powered Script, Voice & Video Engine

SudoVid parses raw production notes and generates fully packaged, ready-to-publish video assets complete with targeted research background frameworks, multi-voice neural audios, and dynamically sequenced video containers.

Requirements

▸ Python 3.11+ build runtime profile.
▸ ffmpeg system layer access (automatically managed via environmental orchestration layers on Cloud deployments).

Local Installation

Clone and setup execution instances locally using standard package managers:

# Clone source repository

git clone https://github.com/sudodocs/sudovid.git

cd sudovid

# Install environment requirements

pip install -r requirements.txt

# Start local server

streamlit run app.py

Deploying to Streamlit Cloud

Ensure your deployment repository root contains these specific orchestration configs:

app.py

requirements.txt

packages.txt

README.md

Note: Your packages.txt file must contain a single entry line reading ffmpeg to force binary installations during deployment environments.

API Credentials Mapping

Enter these verification tokens within the UI tab blocks during data synthesis steps:

Credential Key	Acquisition Origin	Pipeline Consumption Scope
Gemini API Key	aistudio.google.com	Research mapping, Script structures, Metadata generation.
TMDB API Key	themoviedb.org/settings/api	Film backdrops, official promotional art assets, and cast photos.
Pexels API Key	pixels.com/api	Atmospheric stock video assets and general context B-roll clips.

Content Execution Modes

🎬 Film & Series

Leverages official video trailer extracts via yt-dlp pipelines, structural movie art assets, and cast metadata processing vectors.

🔍 Tech News

Prioritizes contextual verified Wikipedia media layers, B-roll structures, and generates sharp impact data visualizer metric frames.

📚 Educational Tech

Outputs syntax-colored development code frames, structural card break timelines, and explanatory visual parameters.

The 6-Tab Processing Loop

Parameters Configuration

Set video slants, lengths, and custom perspectives via tone matrix adjusters.

Targeted Ground Research

Executes deep context lookups with search engines enabled to parse structural source facts.

Script Synthesis Engine

Generates editable conversational markdown scripts comprising clear hooks, acts, and outcomes.

Neural Audio Generation

Transforms scripts into crisp vocals using high-fidelity edge network reading sound nodes.

Metadata Bundle Assembly

Generates high-CTR descriptions, SEO targeting tag layers, and thumbnail prompts.

Compositing & MP4 Renders

Fetches visual layers, builds timed shot blocks with fluid zoom parameters, overlays text scripts, and renders out structural MP4 file containers.

Format Output Canvas Specifications

YouTube Shorts

Dimensions: 1080 × 1920 (9:16 aspect canvas)
Timing Parameters: Under 60 seconds duration limit
Overlays: Pill-encased centralized header titles

Desktop Long-Form

Dimensions: 1280 × 720 (16:9 aspect canvas)
Timing Parameters: Matches exact parsed text runtime
Overlays: Lower-left minimal opacity tags

Known System Boundaries

Execution Safety Parameters

• Keep the web runtime interface tab open continuously while video compilation rendering scripts execute.

• Interface states flush completely during manual browser reload events—always clear tasks cleanly using UI control buttons inside side bars.

• Free deployment execution pipelines possess shared basic computation thresholds, rendering deeper long-form analyses across slower multi-minute pipelines.