{
  "id": 20373654,
  "title": "YouTube's transcript panel changed its DOM: the new selectors that actually work",
  "url": "https://urgent.news/2026/11/17/youtubes-transcript-panel-changed-its-dom-the-new-selectors-that",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-17T03:45:58.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/alapha888/youtubes-transcript-panel-changed-its-dom-the-new-selectors-that-actually-work-4a2n"
  },
  "original_language": "en",
  "account": "While developing a YouTube transcript downloader (a browser extension that I submitted to Edge Add-ons and is currently under review, not yet live), I discovered that the selectors used in older instructions no longer yield any results in the live transcript panel. Here's a snapshot of what the panel looks like as of 2026-11-15 and 2026-11-16. The outdated selector ytd-transcript-segment-renderer, which older guides and code samples targeted, now returns no hits in the live panel.\n\nEach transcript row is now represented by a transcript-segment-view-model element. To extract the data, my tool first attempts to use the new DOM, and if that fails, it falls back to the old selector, ensuring compatibility with both renderings. I tested this on a real video — Rick Astley's Never Gonna Give You Up (dQw4w9WgXcQ). With the transcript panel open, the extension retrieved 24 cues and 2,066 characters from the panel, successfully constructing all four formats: TXT (2,335 bytes), MD (2,668 bytes), SRT (3,142 bytes), and VTT (3,162 bytes). The content was cross-checked and found to be correct. However, there are two important points to consider before relying on the panel:\n\n1. The transcript panel only displays rows currently visible on the screen. In this video's case, it showed only the first 24 cues — not the entire transcript. For the full transcript, you'll still need to use the timedtext endpoint.\n\n2. The timedtext endpoint does not work from a datacenter IP. From my location, I was able to successfully retrieve 0-byte bodies for all formats using a real browser, but attempts through the server returned HTTP 200 with empty bodies and HTTP 429 Too Many Requests responses. It seems that residential IPs might work, but I couldn't verify that situation from here. Therefore, the most reliable order for extracting the transcript is to first try timedtext, then fall back to using the panel's DOM selectors. If you maintain a transcript tool, it's crucial to verify your selectors against the live panel, as the old ones now return nothing.",
  "summary": "While building a YouTube transcript exporter (a browser extension I submitted to Edge Add-ons, currently in review — not live yet), I found that the selectors in older guides return zero hits on the live transcript panel. Here is what the panel actually renders now, measured on 2026-11-15 and 2026-11-16. The old selector is dead ytd-transcript-segment-renderer — the element older guides and code…",
  "key_points": [
    "YouTube transcript panel changed DOM, new selectors work",
    "Transcript rows now represented by transcript-segment-view-model",
    "Use timedtext endpoint for full transcript, fallback to panel"
  ],
  "editors_take": "YouTube's structural updates to the transcript panel force developers to abandon legacy selectors and adopt more complex fallback methods to maintain reliable data extraction from the live DOM.",
  "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."
}