The runbook was 34 pages. Thirty-four. A service migration guide that three different engineers had contributed to over eighteen months, with sections titled things like "Pre-Migration Dependency Audit" and "Rollback Procedure for Partial Cutover Failures." Important stuff. The kind of document that exists specifically because someone got paged at 3am and nobody could find the instructions.
I needed to read it before oncall rotation started on Monday. It was Thursday evening and I had been staring at Jira tickets since 9am.
My eyes were done.
Confluence Has No Read-Aloud Feature
Atlassian has built project management tools, CI/CD pipelines, code review systems, incident management platforms, and a wiki that millions of teams depend on daily. They have not built a way to listen to that wiki.
There's no "Listen" button on Confluence pages. No text-to-speech option in the menu. No accessibility feature that reads content aloud. You can export to PDF and use a separate PDF reader with TTS — which is roughly as convenient as printing a webpage to fax it to yourself.
Chrome has built-in accessibility features. You can enable "Select-to-speak," highlight a section, and hear your OS's default voice read it. The voice sounds like it was recorded in 2013 because it was. It reads one selection at a time. It doesn't track your position. It doesn't understand that the sidebar navigation saying "Quick Links" is not part of the runbook you're trying to learn.
That's not reading documentation. That's fighting your browser to approximate reading documentation.
What I Actually Wanted
Take the Confluence page — all 34 pages of runbook — and read it to me. Natural voice. Show me which paragraph it's on. Let me speed it up because I already know the first six sections. Let me click paragraph 47 to jump directly to the rollback procedures.
Turn Confluence into a podcast episode that I can listen to while cooking dinner.
CastReader does exactly this.
Why Confluence Pages Are Tricky for TTS Tools
Confluence pages look simple — they're wiki pages with headings and paragraphs. But the actual DOM is anything but simple.
A typical Confluence Cloud page has: the top navigation bar with product switcher, the space sidebar with page tree, breadcrumbs, the page title area with author/date metadata, action buttons (Edit, Share, Watch, etc.), the actual page body content, inline comments anchored to specific text, a comment section below, and various Atlassian Connect app panels.
A naive "read everything on the page" approach would produce: "Confluence Home Spaces People Apps Templates Create Service Migration Runbook Engineering Platform Migration Guides Edit Share Watch three dots Last updated April 2026 by Jane Smith This runbook covers the end-to-end process..." — useless.
CastReader extracts only the page body. It identifies the Confluence content area, pulls out headings, paragraphs, lists, table text, and expand block content (if expanded), and ignores everything else. The result is clean, structured text in the right reading order.
Step by Step
Install CastReader. Chrome Web Store. Free. No account, no API key, no trial.
Open any Confluence page. A runbook, an architecture decision record, a meeting notes page, a team handbook, a postmortem — whatever you need to read.
Click the CastReader icon in your Chrome toolbar. Playback starts immediately. A floating player appears at the bottom. Each paragraph highlights as it's read. Click any paragraph to jump there.
The 34-page runbook? Listened to it in 22 minutes at 1.5x while walking home from the office. Showed up to oncall rotation actually knowing the rollback procedures.
Where This Matters Most
Onboarding documentation. New hires have dozens of Confluence pages to get through in their first two weeks. Team handbooks, coding standards, architecture overviews, process guides. Listening to them during commute or lunch turns a dreaded "reading week" into something manageable.
Incident runbooks. The irony of runbooks is that you need them most when you're least able to read carefully — during an incident, stressed, at 2am. Hearing the steps read aloud while your hands are on the keyboard executing commands is genuinely safer than trying to read and type simultaneously.
Architecture Decision Records (ADRs). These are the documents that explain why your system is designed the way it is. They're crucial for understanding context, but they're also dense multi-page essays that nobody reads until they need to. Listening makes them accessible.
Meeting notes and postmortems. Your team writes meeting notes in Confluence. Your team also has too many meetings. Listening to yesterday's meeting notes while doing email triage is multitasking that actually works, because the audio and visual tasks use different cognitive channels.
Compliance and policy documents. Security policies. Data handling procedures. Code review guidelines. SOC 2 documentation. These documents exist because someone must read them. Making them listenable means they actually get consumed instead of sitting in a space that everyone bookmarks and nobody opens.
Works on Cloud, Data Center, and Server
CastReader reads the rendered page in your browser. It doesn't call the Confluence API. It doesn't need admin permissions. It doesn't need a service account or OAuth token.
This means it works on:
- Confluence Cloud (yourcompany.atlassian.net/wiki)
- Confluence Data Center (your-confluence.internal.company.com)
- Confluence Server (legacy self-hosted instances)
Your IT team doesn't need to approve anything. You install a Chrome extension on your own browser. The page content is sent to CastReader's TTS server to generate audio, processed in real-time, and not stored.
Browser TTS vs. CastReader
Your browser's built-in TTS uses system voices — "Samantha" on Mac, "David" on Windows. These are voices from 2014 that pronounce "Kubernetes" as "koo-BER-neh-tees" and have zero understanding of technical documentation rhythm.
CastReader uses neural AI voices trained on natural speech patterns. Technical terms, acronyms, and mixed-case identifiers get handled properly. Long paragraphs don't become monotonous.
Browser TTS also has no awareness of Confluence's page structure. It would read the sidebar, the breadcrumbs, the "Last updated" metadata, and the comments section. CastReader reads only the page body.
Works on Google Docs and Notion Too
Same idea, different platforms. CastReader has dedicated extractors for Google Docs (which uses a canvas-based rendering system that makes extraction non-trivial), Notion (block-based DOM with data attributes), and ChatGPT, Claude, Gemini, and dozens more.
Every platform gets clean extraction without any configuration. CastReader detects the site and picks the right strategy.
The Cost
Free. No word limits. No daily caps. No "premium voices" tier. No team license.
Every other TTS tool I tried had a paywall within three paragraphs. A 34-page Confluence runbook would exhaust a free tier before the table of contents.
CastReader has no free tier because there's no paid tier. It's just free.
Install CastReader and turn your Confluence knowledge base into something you can actually listen to. Your oncall self at 3am will thank you.
