How to Build a $0 Fully Automated Tech Video Pipeline with Open-Source Tools
Introduction
Creating technical video content for platforms like YouTube, LinkedIn, or Twitter usually requires a web of paid SaaS subscriptions: video editing software, text-to-speech APIs, stock media libraries, and paid LLM tokens.
In this guide, we will break down how to build a 100% free ($0-cost), fully automated video production pipeline using open-source tools, free-tier LLM endpoints, and programmatic animation frameworks.
🛑 The Problem
- High API & Subscription Costs: Proprietary AI voice generators (ElevenLabs), video rendering platforms, and paid LLM APIs quickly add up to hundreds of dollars per month.
- Manual Editing Bottlenecks: Traditional GUI video editors (Premiere, DaVinci) require hours of manual timeline tweaking for simple code snippets and architectural diagrams.
- Vendor Lock-in: Cloud video generation platforms restrict customization and force reliance on proprietary servers.
💡 The Solution
A scriptable, headless video pipeline where AI agents handle research, scripting, audio generation, programmatic animation rendering, and video assembly without any manual UI interaction or paid subscriptions.
The $0 Tech Stack
| Pipeline Stage | Tool / Framework | Cost | Official Resources |
|---|---|---|---|
| Research & Trends | GitHub REST API, arXiv API, Papers with Code | $0 | GitHub API Docs |
| Script Synthesis | Free LLMs (Groq, Google AI Studio, Ollama) | $0 | Groq Console |
| Voiceover (TTS) | Kokoro TTS / Edge TTS | $0 | Kokoro GitHub | Kokoro HF |
| Programmatic Visuals | Hyperframes (Apache 2.0) | $0 | Hyperframes GitHub | Official Docs |
| Video Assembly | FFmpeg CLI | $0 | FFmpeg Official |
🛠️ Step-by-Step Installation & Setup
1. Project Prerequisites & requirements.txt
Create a clean Python environment and save the following dependencies:
# requirements.txt
requests>=2.31.0
kokoro-onnx>=0.3.1
soundfile>=0.12.1
numpy>=1.26.0
groq>=0.4.0
Install Python dependencies and FFmpeg:
# Install Python packages
pip install -r requirements.txt
# Install FFmpeg (Linux / macOS / Windows)
# Debian/Ubuntu:
sudo apt update && sudo apt install -y ffmpeg Node.js npm
# macOS:
brew install ffmpeg node
# Verify installations
ffmpeg -version
node -v
2. Installing Hyperframes
Hyperframes is an open-source (Apache 2.0) HTML-to-video rendering engine that lets AI agents write video scenes using web standards (HTML, CSS, JS, GSAP).
# Initialize a new Hyperframes project
npx hyperframes init my-video-project
cd my-video-project
# Install dependencies
npm install
# Test rendering a sample scene locally
npx hyperframes render
3. Setting Up Kokoro TTS
Kokoro-82M is a lightweight, open-weight text-to-speech model (82M parameters) that delivers high-quality audio outputs locally or via ONNX runtime.
# Install Kokoro ONNX package and download lightweight voice weights
pip install kokoro-onnx soundfile
# Download Kokoro ONNX model files (English voice sample)
wget https://github.com/thewhitetulip/kokoro-onnx/releases/download/v0.2.0/kokoro-v0_19.onnx
wget https://github.com/thewhitetulip/kokoro-onnx/releases/download/v0.2.0/voices.json
# Quick Test Script: generate_audio.py
from kokoro_onnx import Kokoro
import soundfile as sf
kokoro = Kokoro("kokoro-v0_19.onnx", "voices.json")
samples, sample_rate = kokoro.create(
"Welcome to this open source automated video pipeline tutorial.",
voice="af_sarah",
speed=1.0,
lang="en-us"
)
sf.write("narration.wav", samples, sample_rate)
print("Saved narration.wav successfully!")
🤖 Full AI Agent Prompt
You can feed this prompt directly to your autonomous AI coding agent (e.g. Hermes, Claude Code, Codex) to execute the pipeline end-to-end:
SYSTEM PROMPT: Autonomous Tech Video Pipeline Agent
Goal: Fetch trending tech topics, generate a short-form video script, synthesize voiceover with Kokoro TTS, generate programmatic code animations with Hyperframes, and assemble the final MP4 using FFmpeg.
Execution Steps:
1. RESEARCH: Query the GitHub REST API for trending repositories in 'Python' or 'AI' over the past 7 days. Select the top repository.
2. SCRIPTING: Generate a 45-second narration script formatted as JSON with timestamps, scene descriptions, code highlights, and speech text.
3. AUDIO: Run 'generate_audio.py' passing the narration text to Kokoro TTS (af_sarah voice) to produce 'narration.wav'.
4. VISUALS: Write HTML/CSS/JS compositions in Hyperframes inside the project directory matching the visual cues, and run 'npx hyperframes render' to output 'visuals.mp4'.
5. ASSEMBLY: Execute FFmpeg to stitch 'visuals.mp4' and 'narration.wav' into 'final_output.mp4':
ffmpeg -i visuals.mp4 -i narration.wav -c:v copy -c:a aac -b:a 192k final_output.mp4
6. VERIFICATION: Ensure final_output.mp4 exists, has non-zero size, and audio/video durations match.
📌 Conclusion & Key Takeaways
By replacing proprietary tools with Hyperframes, Kokoro TTS, and FFmpeg, you can build a resilient, $0-cost content engine completely under your control.
Top comments (2)
The hardest part of an automated video pipeline is usually not stitching the tools together. It is keeping editorial intent visible: why this scene, why this caption, why this pacing. Without that layer, automation just makes generic output faster.
Spot on! Automation without editorial intent just scales noise.
In this setup, we handle that by feeding structured scene-by-scene JSON blueprints (with explicit visual cues, code highlights, and timing constraints) directly to the script synthesis agent. Keeping the prompt tightly bounded to clear narrative beats prevents generic output.