{
  "id": 12084027,
  "title": "Explain Aloud: If the Screen Disappears, Does the Answer Still Make Sense?",
  "url": "https://urgent.news/2026/10/05/explain-aloud-if-the-screen-disappears-does-the-answer-still-make",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-10-05T05:41:46.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/adarsh-singh106/explain-aloud-if-the-screen-disappears-does-the-answer-still-make-sense-5ghd"
  },
  "original_language": "en",
  "account": "Explain Aloud is a Chrome extension designed to enhance the listening experience of ChatGPT responses for users who prefer voice output. The extension works by extracting the structure of completed ChatGPT responses into typed semantic blocks, such as paragraphs, headings, lists, code blocks, and tables. This structure-aware process allows for a more coherent narration of the response content.\n\nThe extension offers two modes: Literal Mode, which takes a deterministic approach for all block types, and Natural Mode, which uses a local Gemma model running in Ollama to generate more listening-oriented phrasing for code and table blocks. The side panel of the extension displays playback controls, narration segments in sequence, and information about the provenance of each segment, indicating whether it came from a rule, the model, or a fallback.\n\nThe core design principle behind Explain Aloud is to ensure that the answer still makes sense even if the screen disappears, maintaining the coherence of the information presented. The extension's demo showcases a voiceover generated with ElevenLabs, with the actual narration processed through Kokoro. The extension attaches itself to completed ChatGPT responses, and the model figures mentioned in the demo are example response content, not measurements of the extension's performance.\n\nThe extension's extraction process is deterministic, meaning it does not involve any model interpretation or guessing. It identifies paragraphs, headings, list items, code blocks, and tables, each carrying its own fields. Code blocks are identified with their respective programming languages, while table blocks are aware of their headers and cell values. A list item's nesting depth is also captured during extraction.\n\nIn Natural Mode, code and table blocks are sent to a local Gemma model through Ollama for phrasing, instead of relying solely on the deterministic parser. The model's output is a candidate narration, which passes through conservative block-specific validation before entering the narration plan. If the validator cannot establish sufficient support, the system falls back to a rule-based transformation of the originally extracted content, ensuring that the final narration remains faithful to the original structure.",
  "summary": "This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend My girlfriend already has a listening workflow. She uses Read Aloud in both ChatGPT and Claude while doing other things — cooking, commuting, winding down after a long day. It works fine for ordinary prose. Then ChatGPT answers a question with a table, or a code block, or a bulleted list with sub-items, and the whole…",
  "key_points": [
    "Explain Aloud enhances ChatGPT responses for voice output users",
    "Two modes: Literal for deterministic, Natural for listening-oriented phrasing",
    "Maintains coherence of information if screen disappears"
  ],
  "editors_take": "This development underscores a shift towards making AI interactions more accessible by ensuring that voice output remains coherent and sensible even without visual context.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}