manez-HF-username/AI-Powered-Blog-Summary-to-Spoken-Audio-Converter

[!CAUTION] Note: The metadata block above is required by Hugging Face Spaces for deployment. I cannot remove this from here because if I do so, the Hugging Face deployment will not work.

🎙️ AI-Powered Blog Summary to Spoken Audio Converter

Turn any blog post into a podcast episode in seconds. Paste a URL, and this app scrapes the content, summarizes it using AI agents, and converts it to natural-sounding speech — all in one click.

🌐 Hosted on Hugging Face Spaces — accessible directly in your browser with no setup required.

🖥️ Usage

1. 1.Open the app in your browser at http://localhost:7860 (or via Hugging Face Spaces)
2. 2.Paste any blog post URL into the Blog URL field
3. 3.Click Generate Podcast
4. 4.Wait a few moments while the AI agents scrape, summarize, and narrate the content
5. 5.Read the generated Blog Summary or listen to the Podcast Audio directly in the browser

https://github.com/user-attachments/assets/2a4b61f7-133e-4123-bcc5-617d077b5a74

📁 Project Structure

├── app.py                  # Gradio UI and main application entry point
├── blog_summarizer.py      # CrewAI agents, tasks, and crew orchestration
├── pyproject.toml          # Project metadata and dependencies
├── requirements.txt        # Pip-compatible dependencies (exported from uv)
├── Dockerfile              # Container configuration
└── .env                    # API keys (not committed to version control)

🚀 Features

• —Automated Web Scraping — Uses Firecrawl to extract clean article content from any blog URL, filtering out ads and navigation clutter
• —Multi-Agent Summarization — A CrewAI pipeline with two specialized agents (a Web Content Researcher and a Content Analyst) that work sequentially to produce high-quality, podcast-ready summaries
• —Text-to-Speech Conversion — Converts summaries to natural audio using the ElevenLabs API
• —Simple Web Interface — A clean Gradio UI that requires zero technical knowledge to use
• —Dockerized — Ships as a Docker container for consistent, reproducible deployment
• —Deployed on Hugging Face Spaces — No installation needed to try the app

🧱 Architecture

    User Input (URL)
          │
          ▼
┌─────────────────────┐
│   Gradio Frontend   │  ← Web UI on port 7860
└─────────┬───────────┘
          │
          ▼
┌─────────────────────┐
│  Blog Scraper Agent │  ← CrewAI Agent + Firecrawl Tool
│  (Web Researcher)   │
└─────────┬───────────┘
          │  Raw blog content
          ▼
┌─────────────────────┐
│ Summarizer Agent    │  ← CrewAI Agent + Gemini 2.5 Flash LLM
│ (Content Analyst)   │
└─────────┬───────────┘
          │  Podcast-ready summary
          ▼
┌─────────────────────┐
│   ElevenLabs TTS    │  ← Converts text to MP3 audio
└─────────┬───────────┘
          │
          ▼
   Summary + Audio Output

🛠️ Tech Stack

📋 Prerequisites

• —Python 3.11
• —Docker (optional, for containerized deployment)
• —API keys for:
• —Google Gemini (GEMINI_API_KEY)
• —Firecrawl (FIRECRAWL_API_KEY)
• —ElevenLabs (ELEVENLABS_API_KEY)

⚙️ Setup & Installation

1. Clone the Repository

git clone https://github.com/your-username/AI-Powered-Blog-Summary-to-Spoken-Audio-Converter.git
cd AI-Powered-Blog-Summary-to-Spoken-Audio-Converter

2. Configure Environment Variables

Create a .env file in the project root:

GEMINI_API_KEY=your_gemini_api_key
FIRECRAWL_API_KEY=your_firecrawl_api_key
ELEVENLABS_API_KEY=your_elevenlabs_api_key

3. Install Dependencies

Dependencies are managed with uv. To export them to a requirements.txt (e.g. for Hugging Face Spaces or Docker), run the following command in PowerShell:

uv export --format requirements-txt | Set-Content -Encoding UTF8 requirements.txt

This generates a clean requirements.txt without hash pins, encoded in UTF-8.

Then install with pip:

pip install -r requirements.txt

Or install directly using uv:

pip install uv
uv sync

▶️ Running the App

Option A: Local

⚠️ Before running locally, update the last line in app.py. Replace: ``python if __name__ == "__main__": demo.launch(server_name="0.0.0.0", server_port=7860, theme="soft") ` with: `python if __name__ == "__main__": demo.launch(theme="soft") ` The servername` and `serverport` settings are intended for Docker/Hugging Face Spaces deployment and are not needed for local use.

python app.py

Then open your browser at http://localhost:7860.

Option B: Docker

# Build the image
docker build -t blog-to-podcast .

# Run the container
docker run -p 7860:7860 --env-file .env blog-to-podcast