How to build an ebook audiobook service
The author produced ebookaloud, an ebook narration product targeting m4b output, and shared a free sample demonstrating its voice and format.
news.ycombinator.comSuggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Build [ebook narration service] using [voice model] and [GPU service]. For each feature, plan, implement, test, review, correct, and commit. Use separate evaluation agents for [extraction pipeline] and [synthesis pipeline]. Produce [audiobook format] and pause for human review of critical code sections.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- flatline
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The author produced ebookaloud, an ebook narration product targeting m4b output, and shared a free sample demonstrating its voice and format.
- Human involvement
- The author read critical generated sections and held extensive conversations with DS Pro about each step.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Plan each change with the coding agents.
- Implement, test, review, correct, and commit each feature.
- Run separate evaluations of extraction and synthesis.
The author built ebookaloud to convert ebooks into audiobooks, using the open Kokoro voice model and a cloud GPU approach after struggling with CPU inference on a laptop. Their goals were to explore multi-agent coding workflows and create a product specifically for ebook narration. They reported that DeepSeek v4 in OpenCode wrote 99% of the application.
Every change or feature followed a plan, implement, test, review, correct, and commit cycle using a mix of Pro and Flash agents. Implementation generally involved one or two concurrent workers. A separate evaluation agent checked parts of the extraction and synthesis pipeline, with eight to ten evaluations running concurrently. The author later set up Claude Code and compared Opus alongside DeepSeek.
The author wrote none of the code but read critical sections and discussed each stage extensively with DS Pro. The product targeted m4b output, pay as you go pricing, and good enough narration quality. A free sample demonstrated the voice and format. Other Kokoro languages and PDF extraction remained possible future additions. Who it's for: Experienced developers willing to supervise agents while building a focused narration product.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · flatline
Frequently asked questions
How did the author supervise generated code?
They read critical sections and discussed each step extensively with DS Pro, although they did not write any code themselves.
What output did the product target?
The author targeted m4b generation for an audiobook experience, with pay as you go pricing and good enough output quality. PDF extraction and other languages were future possibilities.
More details
Remaining steps
- Set up Claude Code and compare Opus with DeepSeek.
About this account
Agents wrote the code and followed a staged development cycle, with separate agents evaluating pipeline quality.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=48661083. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: flatline on Hacker News
Source published
Added to Agents In Use
Source checked
Tools and topics
OpenCode · DeepSeek v4 · Kokoro · Claude Code
Audiobooks · TTS · Agent workflows
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.