returns empty array for transcript

[rgsk]

earlier it was working fine, but since 16th May 2025, this is not working in chrome extension.
it's working fine in nodejs (although sometimes returns empty array, but in 2nd attempt it works).

[rgsk]

I had also embedded a content script in youtube video page (via extension), there also it was working fine, since requests were made from 'www.youtube.com', but since 16th May, that also broke and request returns an empty array.

[cherfaoui-syphax]

I had also embedded a content script in youtube video page (via extension), there also it was working fine, since requests were made from 'www.youtube.com', but since 16th May, that also broke and request returns an empty array.

I'm also trying to use it inside a content script ? did you manage to find a way to get the transcript ?

[juanmiguelguerrero]

I have same issue

[cherfaoui-syphax]

I have same issue

I had also embedded a content script in youtube video page (via extension), there also it was working fine, since requests were made from 'www.youtube.com', but since 16th May, that also broke and request returns an empty array.

hey , i solved the problem by swithing to https://ytjs.dev/, it only works if you call it inside a content script but if you want to you can call that content script via messaging from other pages. by the way you can scrap any video you want and not just the one you are on

import { ClientType, Innertube } from 'youtubei.js/web'

export async function getTranscript(videoId = 'e8krKpuaby8', lang = 'ja') {
  const yt = await Innertube.create({
    client_type: ClientType.WEB,
    lang: lang,
    region: 'JP',
    fetch: async (input, url) => {
      return fetch(input, url)
    },
  })
  let info = await yt.getInfo(videoId, 'WEB')
  let scriptInfo = await info.getTranscript(lang)
  return scriptInfo.transcript.content.body.initial_segments.map((segment) => ({
    text: segment.snippet.text,
    startMs: segment.start_ms,
    endMs: segment.end_ms,
  }))
}

// Original message listener
chrome.runtime.onMessage.addListener((request, sender, sendResponse) => {
  console.log('Received message:', request)
  if (request.action === 'GET_TRANSCRIPT') {
    // Use an async IIFE to handle the async operation
    ;(async () => {
      try {
        const { videoId, lang } = request
        const result = await getTranscript(videoId, lang)
        sendResponse({
          action: 'TRANSCRIPT_RESULT',
          transcript: result,
          requestId: request.requestId,
        })
      } catch (error) {
        sendResponse({
          action: 'TRANSCRIPT_ERROR',
          error: error.message,
          requestId: request.requestId,
        })
      }
    })()
    return true // Keep message port open
  }
})

[rgsk]

I used this package - youtube-transcript-api
it works in nodejs only, and cloud IP is not blocked, serving me really well for now

import TranscriptAPI from "youtube-transcript-api";

const youtubeTranscriptMethod2 = async (
  videoId: string
): Promise<TranscriptResponse[]> => {
  const result = (await TranscriptAPI.getTranscript(videoId)) as {
    start: string;
    text: string;
    duration: string;
  }[];
  const transcript: TranscriptResponse[] = result.map((r) => ({
    text: r.text,
    duration: Number(r.duration),
    offset: Number(r.start),
  }));
  return transcript;
};

[dcsan]

@rgsk that is just using some other companies scraping tool, via the web. pretty sure they'll shut that at some point.

what other packages are ppl using? this has stopped working for me today.

https://github.com/0x6a69616e/youtube-transcript-api/blob/main/src/index.js#L14

edit - yeah i just get this on any video:

IngesterService Error: video unavailable or captions disabled

[danielxceron]

Hi! I've solved this issue

The Problem: JSON parsing fails when extracting captions due to YouTube's HTML structure changes

My Fix: Added a regex fallback in the captions parsing logic:

} catch (e) {
  const altMatch = videoPageBody.match(/"captions":\s*({[^}]+})/);
  if (altMatch) {
      return JSON.parse(altMatch[1]);
  }
  return undefined;
}

[dcsan]

@danielxceron do you mean here?

https://github.com/Kakulukian/youtube-transcript/blob/master/src/index.ts#L102-L104

I guess this whole plugin is just one file so it's pretty easy to patch?

[danielxceron]

Correction - Working Solution Using InnerTube API
Hi,
I apologize for the confusion in my previous comment where I suggested a solution that wasn't fully functional. I was too hasty in commenting without properly testing the implementation.
After deeper research, I've developed a working solution that uses YouTube's InnerTube API, which is more robust and reliable than the previous HTML parsing method.