ShowAI

A blazing fast, programmatic browser voiceover & timeline generator.

ShowAI is a Python framework that completely automates the creation of narrated browser recordings. It allows you to programmatically define a "timeline" consisting of browser actions, wait periods, and AI-generated voiceovers. The engine runs Playwright under the hood, plays the generated voice in sync with the actions, captures the raw video, and seamlessly stitches everything together using FFmpeg.

---

🚀 Features

• Programmatic Timeline: Chain add_action(lambda page: ...), add_wait(sec), and add_voice("Text") to script your perfect demo.
• Dual TTS Engine Architecture:
  • 🎙️ F5-TTS: High-quality, zero-shot voice cloning directly from your .wav/.m4a files.
• Cinematic Execution: Features a built-in boot-up sequence, terminal ASCII art banner, and a professional video splash screen using your logo.png.
• Auto-Stitching: Transparently delegates video + audio track alignment to FFmpeg, guaranteeing perfectly synced .mp4 outputs.

---

🛠️ Installation

ShowAI is built for modern Python (3.10+). We recommend using uv for lightning-fast dependency resolution.

# 1. Create your isolated environment
uv venv --python 3.12 venv
source venv/bin/activate

# 2. Install ShowAI in editable mode
uv pip install -e .

# 3. Ensure browser drivers are present
playwright install chromium

---

💻 Quickstart

A robust example.py is included in the project directory. It demonstrates how to rapidly spin up a voice-synced walkthrough using F5-TTSVoice Cloning.

import os
from dotenv import load_dotenv

from showai import ShowAI, F5TTSEngine

# Define Voice Engine (Using your own zero-shot profile)
engine = F5TTSEngine(ref_audio="lgvoice.wav")

app = ShowAI(tts=engine, output_video="demo1.mp4")

# 1. Introduce the video
app.add_voice("Welcome to my ShowAI project.")

# 2. Add visual interactions via Playwright
app.add_action(lambda page: page.goto("https://tonai.luisguilher.me"))
app.add_wait(1.5)

# 3. Speak while navigating
app.add_voice("The timeline engine perfectly syncs my voice to these exact moments.")
app.add_wait(2.0)

if __name__ == "__main__":
    app.execute()

Run your timeline from your terminal (the ASCII banner will welcome you!):

python example.py

---

🧠 Architecture Setup

Timeline Orchestrator (ShowAI)

The Orchestrator caches the entire timeline before spinning up the browser. All heavy TTS generation happens before the browser window opens to ensure perfectly buttery-smooth Playwright recordings.

Browser Automation (BrowserAutomation)

Runs in headless=False so you visually confirm the script is working. It automatically renders a cinematic center-framed black HTML screen containing the logo for 0.5s before kicking off your goto and click logic.

Missing FFmpeg?

If the script crashes at the final video stitching step, please ensure your system has FFmpeg installed:

• Ubuntu/Debian: sudo apt install ffmpeg
• macOS: brew install ffmpeg

Built to automate the un-automatable.