How to Listen to ChatGPT Responses
• By Elliott Tong
To listen to ChatGPT responses, install the Alexandria Chrome extension, then open any ChatGPT conversation. Alexandria detects when a response finishes streaming and injects a play button into the action toolbar beside the copy button. Click it to hear that response read aloud with word-by-word highlighting. Each response in a conversation gets its own button, so you can listen to any one without scrolling back to find it.
Why Listen to AI Responses Instead of Reading Them?
Long AI responses are a new reading problem. A single ChatGPT answer can run thousands of words — longer than many articles. After a few back-and-forth turns, the conversation becomes a wall of text that demands the same focused reading as a document, but without the structure.
Listening solves this in a few ways:
You can multitask
Listen while coding, sketching, or doing the task ChatGPT is helping with. The response becomes background audio rather than something that pulls you out of your flow.
It reduces eye strain
Programmers and knowledge workers already spend hours reading code and documentation. Adding long AI responses on top compounds the strain. Listening lets your eyes rest while you still absorb the content.
You catch more detail
Speed-reading through AI responses means skipping sections that look similar to what you've already read. Listening forces you to go through the whole thing, and dual coding (hearing words while seeing them highlighted) improves retention over reading alone [Paivio, University of Western Ontario; Lindamood-Bell Learning Processes].
You can compare responses
When you've asked ChatGPT to regenerate or refine an answer, listening to both versions back-to-back is faster than scanning them visually and trying to spot the differences.
How Alexandria Works with ChatGPT
Alexandria's ChatGPT adapter (584 lines of custom code) handles three challenges that generic TTS tools cannot:
Streaming-aware injection
ChatGPT streams responses token by token. A naive extension would inject a button into a half-finished response, then break when the DOM shifts as more tokens arrive. Alexandria's adapter watches for streaming completion indicators (cursor elements, streaming attributes, completion CSS classes) and only shows the play button once the response is stable [ChatGPTAdapter.js, isStreaming()].
Per-response buttons
Each AI response in a conversation gets its own play button, injected into the action toolbar (the row that contains Copy, Good Response, Bad Response, and Share). Click any button to listen to that specific response. This is different from page-level TTS tools that read the entire page from top to bottom.
Multi-response state synchronisation
When you click a response's play button, that button shows the playing state while all other response buttons in the conversation show idle. If you click a different response mid-playback, the first stops and the second starts — no overlapping audio, no confused states [ChatGPTAdapter.js, responseButtons Map + _activeResponseElement tracking].
Conversation switch detection
ChatGPT is a single-page application. Clicking a different conversation in the sidebar doesn't trigger a page load — it just swaps the DOM. Alexandria detects these navigation events via URL polling, cleans up the old conversation's buttons, and injects new ones once the new conversation renders.
Tips for Listening to ChatGPT Effectively
Start with long explanatory responses
ChatGPT's best use case for listening is when you've asked it to explain something in depth — a concept, a codebase, a decision tree. These multi-paragraph responses are where listening saves the most time compared to reading.
Speed up gradually
At 1x, a 500-word ChatGPT response takes about 4 minutes to hear. At 1.5x, it's under 3 minutes. At 2x, it's 2 minutes. Alexandria's free tier goes up to 5x, but start at 1.25x and increase as your comprehension adjusts. Most people settle into 1.5x–2x within a week.
Let highlighting guide you
Word-by-word highlighting keeps you anchored even when you look away. Glance back at any point and you'll know exactly where the audio is. This is especially useful for code explanations where you want to look at the code block while hearing the description.
Use it for regenerated answers
When you ask ChatGPT to try again or refine its answer, you end up with multiple similar-looking responses. Listening to them in sequence is faster than scanning for differences, and the audio makes it easier to notice subtle changes in phrasing.