deepdotspace/podcastify

Podcastify is a web application that converts articles, PDFs, or pasted text into a two-host AI podcast you can stream and download. You pick two fictional hosts, choose a tone or vibe for the conversation, set a length target, and click Generate. The app then runs everything in the background so you can close the tab and come back later. The conversion happens in four steps. First, the app fetches the source content: it can scrape a web page using a tool called Firecrawl, pull text from a PDF using CloudConvert, or just take whatever you paste in. Second, an AI model turns that content into a natural two-person dialogue complete with a title, written in the vibe you chose. Third, each line of the dialogue gets turned into speech using ElevenLabs, with a distinct voice for each host. Fourth, all those audio clips get stitched together into a single MP3 file, stored in cloud storage, and saved to your personal library along with a transcript that stays in sync as the audio plays. The technical foundation is the DeepSpace SDK, which handles user accounts, access control, background job processing, and the cloud storage layer. The front end is built with React, Vite, and Tailwind, following a component library style. The app runs on Cloudflare Workers and Durable Objects, which are serverless compute services that keep costs low and scale automatically. To run your own copy, you need Node and npm installed. After cloning, you run npm install, log in with the DeepSpace CLI, and then use commands to start a local dev server, run the automated test suite, or deploy to your own subdomain under app.space. The costs for third-party services including ElevenLabs, CloudConvert, and the AI provider are paid by the developer who deploys the app, which keeps it free for end users. The live version runs at podcastify.app.space. The codebase is organized under a src/ directory with separate folders for schemas, job definitions, pages, and UI components, plus a worker file for the Cloudflare Worker logic.