I asked ChatGPT to explain how TCP congestion control works. It gave me 1,400 words. Fourteen hundred. With code examples, numbered lists, three analogies involving water pipes, and a section titled "Putting It All Together" that was longer than most blog posts I've written.
I stared at it for about nine seconds before my eyes glazed over.
This is the weird paradox of ChatGPT in 2026. The answers are genuinely good. Often great. But they're long. And reading a 1,400-word explanation on a screen — after you've already been staring at a screen to type the prompt — is not how anyone wants to absorb information at 11pm on a Tuesday.
I wanted to listen to it. Like a podcast. While making dinner.
ChatGPT Has Voice Mode. Why Not Just Use That?
Because voice mode is a conversation, not a reader.
ChatGPT's voice feature (the one where you tap the headphone icon on mobile) is designed for back-and-forth dialogue. You speak, it responds, you interrupt, it adjusts. It's impressive tech. But if you've already typed a question and gotten a long written response, voice mode won't read that response to you. It starts a new conversation. Completely separate context.
There's a small "read aloud" button on each message. It uses the browser's built-in speech synthesis. You know what that sounds like. Flat. Robotic. Every word at the same cadence, like a GPS navigator reciting Shakespeare. The voice stumbles on code blocks. It reads bullet points as one continuous sentence. It pronounces "nginx" as "en-jinx." After thirty seconds you stop processing the content and start processing how much the voice annoys you.
That's not listening. That's punishment.
What I Actually Wanted
Simple: take whatever ChatGPT just wrote, read it aloud with a natural-sounding voice, highlight each paragraph as it goes, and let me pause/resume. Basically turn any ChatGPT response into a mini podcast episode.
No copy-pasting into another app. No downloading the text. No signing up for anything.
CastReader does exactly this. One click.
How It Works on ChatGPT
Here's the thing about ChatGPT's page structure — it's not a normal website. The DOM is a React app with dynamically rendered conversation threads, nested markdown rendering, code blocks with syntax highlighting, copy buttons, thumbs up/down widgets, and model selector dropdowns all mixed in with the actual content you care about.
A generic text extractor would grab all of it. The navigation. The "ChatGPT can make mistakes" footer. The model name. The timestamp. Every piece of UI text alongside the actual answer.
CastReader doesn't do that. It has a specialized extractor built specifically for LLM platforms — ChatGPT, Claude, Gemini, and others. The extractor understands the conversation structure. It identifies assistant messages, strips out the UI chrome, preserves paragraph boundaries and code blocks, and extracts just the content you'd want to hear.
The voice is natural AI speech. Not the browser's built-in engine. Actual neural TTS that handles pauses, emphasis, and sentence rhythm the way a human reader would.
Step by Step
Install it. Go to the Chrome Web Store and add CastReader. Zero signup. No account. No trial period. It's free.
Ask ChatGPT something. Anything that produces a long response. "Explain the difference between SQL and NoSQL databases." "Write me a 500-word essay about urban planning in Tokyo." "How does mRNA vaccine technology work?" The meatier the better.
Click the CastReader icon in your Chrome toolbar (or press the keyboard shortcut). That's it. Playback starts immediately. Each paragraph highlights as it's read. A floating player bar appears at the bottom — pause, skip forward, adjust speed.
You can click any paragraph to jump directly to it. Already heard the intro? Click paragraph four. Want to re-listen to that tricky explanation about database indexing? Click it again.
Where This Actually Matters
Code explanations. ChatGPT writes surprisingly good code walkthroughs, but reading a 20-paragraph explanation of how a recursive algorithm works, while simultaneously looking at the code, is cognitive overload. Listening to the explanation while looking at the code is dramatically better. Your eyes stay on the code. Your ears get the narrative. Two input channels instead of one.
Research rabbit holes. You asked about the history of the Suez Canal. ChatGPT gave you 900 words covering the 1869 construction, the 1956 crisis, and the 2021 Ever Given incident. You don't want to read all of that. But you'd happily listen to it while walking your dog.
Essay drafts. You asked ChatGPT to draft something for you. Now you need to evaluate whether it's any good. Reading your own writing is hard enough — reading AI-generated text critically is harder because it all looks polished. Hearing it read aloud exposes the awkward transitions, the repetitive phrasing, the paragraphs that say nothing in fifty words.
Accessibility. Some people process audio better than text. Some people have eye strain. Some people just spent eight hours reading documents at work and cannot stare at one more screen. This isn't a niche use case — it's most of us by 6pm.
Browser TTS vs. CastReader
Your browser has a built-in speech engine. Chrome uses whatever system voices your OS provides. On a Mac that's "Samantha." On Windows it's "David" or "Zira." They sound like computers from 2014 because that's roughly when those voice models were trained.
CastReader uses AI-generated voices — neural text-to-speech trained on natural speech patterns. The difference isn't subtle. It's the gap between a MIDI piano and a Steinway grand. One is technically playing the notes. The other is making music.
Also: browser TTS has no paragraph tracking, no highlighting, no player controls, and no understanding of page structure. It just reads everything it finds, including "Copy code" buttons and navigation menus. CastReader's specialized LLM extractor skips all that noise.
For a deeper comparison, see our breakdown of free TTS tools.
Works on Claude and Gemini Too
Same specialized extraction. Same one-click experience.
Claude's responses tend to be structured differently from ChatGPT's — more headers, longer paragraphs, different markdown conventions. Gemini has its own DOM structure entirely. CastReader handles all three because each platform has a dedicated extractor that understands its specific page layout — just like it has dedicated extractors for Notion pages and Medium articles. You don't configure anything. It detects the site and picks the right extraction strategy automatically.
The Cost
Free. Completely. No word limits. No daily caps. No "premium voice" upsell. No account creation.
I keep mentioning this because every other TTS tool I tried before CastReader had some version of "free for 500 characters, then $9.99/month." Five hundred characters is about three sentences. That's not free — that's a demo.
CastReader has no demo. It's the whole thing, for nothing.
Install CastReader from the Chrome Web Store and turn ChatGPT into something you can listen to. Your eyes will thank you.
