IdeaWell

Project ideas from Hacker News discussions.

Show HN: Use your OpenAI subscription to create audiobooks

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

Theme 1 – Misleading title / confusion about subscription vs. API key
Many commenters felt the claim “Use your OpenAI subscription to create audiobooks” was deceptive because the tool still requires an external API key for the text‑to‑speech step.

  • “I thought this would mean I could use my subscription… Instead this asks for an API key, which is very different from ‘Use your OpenAI subscription’.” — embedding-shape
  • “The title is misleading - should be 'Use your openai subscription AND a TTS api key'.” — scrollop
  • “Your subscription will be used for: document processing (via MCP)… Your API key will be used for: generating speech OpenAI charges per usage.” — wastewastewaste

Theme 2 – Security/privacy concerns about pasting API keys to third‑party sites
Several users warned against entering personal API keys on unfamiliar web services and advocated for local or self‑hosted alternatives.

  • “You don't even need that. Just paste your pdf in codex or the gpt work app and say make me an audio book…” — fnord77
  • “There are standard practices for delegating responsibility, never ever be pasting API keys into websites.” — xnorswap
  • “For those who do not want to paste OpenAIkeys to other peoples web pages, there are local solutions… runs fully offline.” — draugadrotten

Theme 3 – Interest in practical features: samples, voice quality, format handling, and workflow
Discussion turned to whether the output sounded good, how it handled complex PDFs, multilingual support, and how to integrate the result into podcast pipelines.

  • “Here you can listen to examples: https://pdftomp3.com/examples/” — janpmz
  • “I like the voices, will be using the product soon.” — brunorsini
  • “How well does it handle larger files? And other book formats, like epub or mobi?” — matthorse
  • “I did something similar… and just put the files on an old laptop and used Tailscale. That gives me an https URL and Apple Podcasts downloads the files fine.” — cj00
  • “What about license of such output? Won't Google and OpenAI be coming to my door for their share…” — butz (followed by advice to check TTS provider licenses).

🚀 Project Ideas

Generating project ideas…

KeyVault PDF-to-Audio: Secure OAuth + Local TTS

Summary

  • Lets users convert PDFs/EPUBs to audio using their own OpenAI/Claude subscription via OAuth device flow, eliminating the need to paste API keys into untrusted sites.
  • Core value proposition: Secure, privacy‑first conversion that keeps credentials under user control while still leveraging powerful LLMs for document understanding.

Details

Key Value
Target Audience Professionals, researchers, students who have LLM subscriptions but dislike sharing API keys
Core Feature OAuth‑based authentication for document processing + optional encrypted local storage of TTS API keys (or use open‑source TTS models)
Tech Stack Frontend: React/Vite; Backend: Node.js (Express) or Python FastAPI; Auth: OpenID Connect device flow; Storage: IndexedDB (encrypted) or local .env; TTS: Web Speech API or fetch to user‑provided endpoint
Difficulty Medium
Monetization Hobby

Notes

  • Addresses the concern raised by draugadrotten: “For those who do not want to paste OpenAI keys to other peoples web pages …”.
  • Enables a self‑hosted or client‑side workflow that HN commenters would trust, encouraging discussion on secure auth patterns for AI‑powered tools.

PodcastFeed Auto: Watch‑folder → Private RSS with TTS Choice

Summary

  • Monitors a user‑chosen folder (Syncthing, Dropbox, local drive) for new documents, converts them to MP3 using the user’s preferred TTS provider, and publishes a private RSS feed ready for Apple Podcasts/Overcast.
  • Core value proposition: Fully automated “drop‑in‑folder → podcast” pipeline that solves the manual copy‑paste hassle and lets listeners choose any TTS voice.

Details

Key Value
Target Audience Power users who listen to articles/papers while commuting and already use file‑sync tools
Core Feature Folder watcher → text extraction → TTS (OpenAI, ElevenLabs, local models) → MP3 → RSS feed with enclosure URLs
Tech Stack Python (watchdog, pypdf/pdfminer, TTS adapters, feedparser); optional Docker; can run on a Raspberry Pi or VPS
Difficulty Medium
Monetization Hobby

Notes

  • Mirrors the workflow praised by robot_jesus: “I did something similar but didn’t do the last mile with the cloud bucket and RSS stream and that seems key.”
  • Provides a concrete basis for discussion on integrating personal TTS keys vs. free local models, and on delivering audio via familiar podcast apps.

DocuSpeak: Layout‑aware PDF‑to‑Audio Library with SSML Pronunciation Control

Summary

  • Open‑source Python library and CLI that accurately extracts text from complex PDFs (tables, multi‑column, footnotes) and feeds it to any TTS engine while supporting SSML for pronunciation, speaker turns, and language switching.
  • Core value proposition: Gives developers fine‑grained control over audio quality and multilingual, multi‑speaker output, solving the formatting and pronunciation pain points.

Details

Key Value
Target Audience Developers building audiobook apps, researchers needing high‑fidelity conversion of academic PDFs
Core Feature Layout‑aware extraction (pdfminer.six + heuristics) → LLM‑guided cleanup → TTS adapters (OpenAI, ElevenLabs, Coqui, F5‑TTS) with SSML support
Tech Stack Python 3.11+, pdfminer.six, langchain or litellm for LLM prompting, ONNX Runtime or torch for local TTS, Click/typer for CLI
Difficulty High
Monetization Hobby

Notes

  • Responds to cj00’s worry about pronunciation (“How well does it handle pronunciation, especially of things like proper nouns?”) and abello’s comment on handling tables/captions.
  • Offers a rich foundation for HN discussion on open‑source TTS pipelines, SSML standards, and community‑driven improvements for academic document consumption.

Read Later