How jatin-mehra119/ai-powered-podcast-to-blog-generator works
Project Type and Purpose
• This is a full-stack AI-powered content generation application with both a web-based interface and CLI.
• Purpose: Convert podcast audio into blog posts, SEO metadata, FAQs, social posts, newsletters, and quotes by orchestrating a transcription service (Whisper), an LLM API (Groq), and optional web research (Tavily).
File-Structure Analysis
• Top-Level Files:
• api.py – FastAPI application entry point
• main.py – CLI entry point
• requirements.txt, Dockerfile, .github/workflows – deployment/configuration
• Directories:
• frontend/ – static HTML, CSS, JS for drag-and-drop UI
• modules/ – core processing logic split into:
– preprocessing.py (audio cleanup + Whisper transcription)
– models.py (abstractions for LLM calls)
– content_generation.py (SEO, FAQ, quotes, social, newsletter)
– generate_blog.py (blog post builder)
• (README mentions output/, temp/, logs/ but they are created at runtime)
Key Architecture Insights from README
• Asynchronous processing and real-time progress updates
• Two modes of invocation: HTTP API + web UI, and CLI
• External dependencies: Whisper Large v3 Turbo, Groq API, optional Tavily API
1. User Interfaces
– Web Browser (frontend: index.html, script.js, styles.css)
– Command-Line Client (main.py)
2. API Layer
– FastAPI Server (api.py)
3. Processing Pipeline (modules/)
– Preprocessor (transcription via Whisper)
– Content Generators (blog, SEO, FAQ, social, newsletter, quotes)
– Model Adapters (Groq API calls, optional Tavily lookup)
4. Storage
– Temporary Storage (temp/)
– Output Storage (output/)
– Logs (logs/)
5. External Services
– Whisper Large v3 Turbo
– Groq LLM API
– Optional Tavily Web-Research API
1. User → API Layer: upload audio + options
2. API Layer → Temp Storage: save uploaded file
3. API Layer → Preprocessor → Whisper Service → transcript
4. Preprocessor → Processing Pipeline
5. Pipeline → Groq API (for text generation)
6. Pipeline → Tavily API (optional research)
7. Pipeline → Output Storage (generated assets)
8. API Layer → User: real-time status + download links
9. CLI Client → modules (same pipeline) → local file output
c. Architectural Patterns and Principles:
• Layered Architecture: UI, application/API, domain logic, infrastructure
• Modularization: separate modules for concerns (transcription vs. blog vs. SEO)
• Pluggable Adapters: model.py abstracts LLM & research APIs
• Asynchronous/Background Processing: avoid blocking HTTP requests for long audio jobs
d. Relevant Technologies/Frameworks (annotate each component):
• FastAPI (HTTP server, background tasks)
• Uvicorn (ASGI server)
• Whisper Large v3 Turbo (audio transcription)
• Groq API (LLM for content generation)
• Tavily API (web research)
• HTML/CSS/JavaScript (frontend)
• Python 3.9+, Pydantic (data validation), asyncio (concurrency)