I became an active audiobooks listener last year, but I found that there wasn't a good app for listening to content that has no audio version available. Then OpenAI's text-to-speech model, which generates hyper-realistic audio 10x cheaper than other products did before, and I saw an opportunity to create something useful.
What’s under the hood?
- Calibre ebook-convert to convert books into markdown
- Vue/Nuxt.js full-stack app
- A lot of node.js scripting for text cleaning and structuring
- Google Cloud Storage for mp3s (really like their SDK for node by the way, recommend)
- shadcn UI library port for Vue (http://shadcn-vue.com)
- self-hosted on VPS
The pricing model: $2 / hour. No subscription or hidden fees. The idea is to charge a 100% markup over OpenAI's prices to align incentives — I want to help you listen to as much content as possible. If OpenAI cuts their prices, I'll cut mine too. I'm open to feedback on pricing, though. As a solo founder, my goal is to turn this into a business so I can quit my job and work on it full-time.
I want to create a free, public library of audio content in the public domain. I'm starting with PG's essays, Jeff Bezos's shareholder letters, “My life and work” by Henry Ford, and “Shape Up” by Basecamp. I plan to add features for users to contribute to the public library by sharing the content they've listened to, as well as features for authors to generate audio for their own content.
Plans for future / what’s on my mind right now:
- Offline mode, download mp3
- Add features for authors to let their audience listen to their books/blog posts. One author already reached out to me and asked for it.
- Integrate with Project Gutenberg to create 70k audiobooks? I don’t know.
- Maybe better text reader like Readwise’s — to switch between reading and listening.
- Maybe make it standalone app usable with your models/OpenAI token, with single payment.
- Bookmarks, notes, quotes. Share your quote as a video snippet with subtitles.
I'd love to hear your feedback and suggestions! I'm excited to ship feature requests as quickly as possible, even within hours or minutes if I can.
