📰 ➡️ 🎙️ Blog to Podcast

A Streamlit app that turns any public blog post into a podcast episode. It scrapes the article with Firecrawl, summarizes it into a conversational script with Azure OpenAI (via an agno agent), and narrates it with ElevenLabs.

🏗️ Repository Structure

├── .github/
│   └── workflows/
│       ├── ci.yml                    # Lint, test, build & push image to ACR
│       └── dependabot.yml            # Dependency updates configuration
├── src/
│   └── blog_to_podcast/
│       ├── app.py                    # Streamlit UI
│       ├── __main__.py               # CLI entry point (`blog-to-podcast`)
│       ├── config.py                 # Settings resolved from env / sidebar
│       ├── summarizer.py             # Firecrawl scrape + Azure OpenAI summary
│       └── tts.py                    # ElevenLabs text-to-speech
├── tests/                            # pytest suite mirroring src/
├── Dockerfile                        # Multi-stage uv + Python 3.12 image
├── pyproject.toml                    # Project, ruff, and pytest config
└── uv.lock                           # Locked dependencies

🚀 Features

• Blog scraping — pulls the full content of any public blog URL via Firecrawl.
• Summary generation — an agno agent produces an engaging summary (≤ 2000 characters).
• Podcast generation — converts the summary to MP3 audio with an ElevenLabs voice.
• Flexible credentials — read from environment variables, overridable in the sidebar.
• Container ready — multi-stage Docker build with a health check and non-root user.

🛠️ Technology Stack

• Language: Python 3.12
• UI: Streamlit
• Agent framework: agno + Azure OpenAI (v1 API surface)
• Scraping: Firecrawl
• TTS: ElevenLabs
• Tooling: uv, ruff, pytest
• CI/CD: GitHub Actions → Azure Container Registry

🏁 Quick Start

Prerequisites

• Python 3.12+
• uv:

  # macOS / Linux
  curl -LsSf https://astral.sh/uv/install.sh | sh
  
  # Windows
  powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

• API keys for Azure OpenAI, ElevenLabs, and Firecrawl.

Local development

uv sync                      # create the virtualenv and install dependencies
cp .env.example .env         # optional: pre-fill credentials
uv run streamlit run src/blog_to_podcast/app.py

Then open http://localhost:8501, fill in any missing keys in the sidebar, paste a blog URL, and click 🎙️ Generate Podcast.

Configuration

Every value can be supplied via environment variable or entered in the sidebar at runtime.

Variable	Required	Description
AZURE_OPENAI_BASE_URL	✅	v1 base URL, e.g. https://<resource>.openai.azure.com/openai/v1/
AZURE_OPENAI_DEPLOYMENT	✅	Chat model deployment name, e.g. gpt-4o
AZURE_OPENAI_API_KEY	✅	Azure OpenAI API key
ELEVENLABS_API_KEY	✅	ElevenLabs API key
FIRECRAWL_API_KEY	✅	Firecrawl API key
ELEVENLABS_VOICE_ID	—	Voice override (default JBFqnCBsd6RMkjVDRZzb)
ELEVENLABS_MODEL_ID	—	TTS model override (default eleven_multilingual_v2)

Tests and linting

uv run pytest
uv run ruff check .
uv run ruff format --check .

Docker

docker build -t blog-to-podcast .
docker run --rm -p 8501:8501 --env-file .env blog-to-podcast

🙏 Credits

Adapted from the ai_blog_to_podcast_agent starter in Shubhamsaboo/awesome-llm-apps, using the uv + Azure OpenAI variant from PR #1096.

📄 License

See LICENSE.