Audify — Document-to-Podcast Generation AI
Audify is an intelligent document-to-audio platform that converts PDF, DOC, and DOCX files into editable podcast scripts and downloadable MP3 episodes.
Developed as an open-source blueprint under the Cloud2 Labs Innovation Hub, Audify demonstrates how document extraction, local small language models via Ollama, and AI voice synthesis can be combined into a practical end-to-end workflow.
It showcases a production-style microservices architecture for extracting text (with OCR fallback for scanned PDFs), generating two-speaker podcast dialogue, and producing polished audio for fast content repurposing.
What It Demonstrates
Audify illustrates how to:
- Process long-form documents without manual script drafting
- Let users choose podcast style before generation (Deep Dive, Storyteller, Debate, Quick Brief, ELI5)
- Let users set jargon level before generation (Low, Moderate, High) to shape audience complexity
- Generate script drafts using a local small language model through Ollama, with optional fallback to openAI
- Keep output grounded in uploaded document context
- Support rapid iteration through script editing before final audio rendering
Designed for developers, content teams, educators, and innovation groups, Audify serves as a reference implementation for AI-driven document-to-audio transformation.
Key Capabilities
Pre-Generation Script Controls
Before script generation, users select the podcast style and jargon level they want, so the script can be tailored to format and audience complexity.
Local LLM Script Generation (Ollama)
Uses a local small language model through Ollama as the primary script generation path.
Document-Aware Script Pipeline
Builds host-and-guest dialogue directly from uploaded document content.
Script Review and
Edit Step
Provides an editor for refining the generated script before audio synthesis.
AI Voice Casting and Synthesis
Supports host/guest voice selection and generates complete podcast audio with downloadable output.
Containerized Microservices Blueprint
Uses Docker-based services for reproducible local development and experimentation.