YouTube Transcript Scraper Pro

Extract clean transcripts from one or many YouTube videos in a single run.

Fastest first run

{
  "videos": ["dQw4w9WgXcQ"],
  "type": "plain",
  "maxConcurrency": 1,
  "maxRetries": 1
}

Validate one known-caption video first, then increase the batch size. Successful transcript rows and unbilled error rows are separated so downstream workflows can retry only unavailable inputs.

This Actor accepts YouTube video URLs or video IDs and returns transcripts in the format you need: plain text, timestamped text, SRT, VTT, JSON, CSV, or all formats at once. It prefers manually created captions when they are available and otherwise uses an auto-generated transcript.

What you get

• Bulk processing for multiple videos
• Support for standard YouTube URLs, short URLs, Shorts URLs, embed URLs, live URLs, and raw video IDs
• Automatically returns the best available transcript, preferring manually created captions when available
• Optional language setting for users who need a specific transcript language
• Structured dataset output for downstream automations
• Output format control for different workflows (LLM prep, subtitle editing, QA, analytics)
• Apify Proxy support for cloud runs where YouTube blocks datacenter IPs

Input

videos (required)

List of YouTube videos.

Supported values:

• Video ID: VIDEO_ID
• Watch URL: https://www.youtube.com/watch?v=VIDEO_ID
• Short URL: https://youtu.be/VIDEO_ID
• Shorts URL: https://www.youtube.com/shorts/VIDEO_ID
• Embed URL: https://www.youtube.com/embed/VIDEO_ID
• Live URL: https://www.youtube.com/live/VIDEO_ID

type (optional)

Output format for transcript.

Allowed values:

• plain (default)
• timestamped
• srt
• vtt
• json
• csv
• all

language (optional)

Leave empty to return the best available transcript in any language. If you enter a language code such as en, es, or de, the Actor will try to return that language and otherwise report which languages are available.

keepMusicMarkers (optional)

• Type: boolean
• Default: false
• When false, markers like [Music] are removed.

dropEmpty (optional)

• Type: boolean
• Default: true
• Removes empty transcript segments after cleanup.

maxConcurrency (optional)

• Type: integer
• Default: 3
• Range: 1 to 10
• Controls how many videos are processed in parallel.

maxRetries (optional)

• Type: integer
• Default: 2
• Range: 0 to 5
• Retries transient transcript request failures.

proxyConfiguration (optional)

• Type: object
• Default: Apify Proxy with the RESIDENTIAL group
• YouTube often blocks cloud datacenter IPs, so proxy use is recommended for Apify platform runs. You can disable it for local/direct runs if transcript requests work without a proxy.

Output

Each dataset item contains:

• recordType (transcript or error)
• success (boolean)
• originalInput (string)
• videoId (string)
• videoUrl (string)
• languageCode (string or null)
• languageName (string or null)
• isGenerated (boolean or null)
• isTranslatable (boolean or null)
• outputFormat (string)
• transcript (string, array, object, or null depending on type)
• segments (array or null)
• segmentCount (integer)
• availableLanguages (array)
• errorMessage (string or null)
• scrapedAt (UTC timestamp)

If a video fails, the item is still returned with:

• success: false
• transcript: ""
• errorMessage explaining the failure

When this Actor is configured for pay-per-event pricing, successful transcript rows are the only billable dataset items. Failed inputs and spending-limit diagnostics are written to the separate Unbilled errors dataset.

Example input

{
  "videos": [
    "https://www.youtube.com/watch?v=VIDEO_ID_1",
    "https://youtu.be/VIDEO_ID_2",
    "VIDEO_ID_3"
  ],
  "type": "all",
  "language": "",
  "keepMusicMarkers": false,
  "dropEmpty": true,
  "maxConcurrency": 4,
  "maxRetries": 2,
  "proxyConfiguration": {
    "useApifyProxy": true,
    "apifyProxyGroups": ["RESIDENTIAL"]
  }
}

Example output item

{
  "recordType": "transcript",
  "success": true,
  "originalInput": "https://www.youtube.com/watch?v=VIDEO_ID_1",
  "videoId": "VIDEO_ID_1",
  "videoUrl": "https://www.youtube.com/watch?v=VIDEO_ID_1",
  "languageCode": "en",
  "languageName": "English",
  "isGenerated": false,
  "isTranslatable": true,
  "outputFormat": "all",
  "transcript": {
    "plain": "Never gonna give you up...",
    "timestamped": "00:00:01.200 Never gonna give you up...",
    "srt": "1\\n00:00:01,200 --> 00:00:03,800\\nNever gonna give you up...",
    "vtt": "WEBVTT\\n\\n00:00:01.200 --> 00:00:03.800\\nNever gonna give you up...",
    "json": [
      {
        "text": "Never gonna give you up...",
        "start": 1.2,
        "duration": 2.6
      }
    ],
    "csv": "start,duration,text\\n1.2,2.6,\"Never gonna give you up...\""
  },
  "segmentCount": 42,
  "errorMessage": null,
  "scrapedAt": "2026-05-22T12:34:56Z"
}

Typical use cases

• Build subtitle files (srt or vtt) for editing or publishing
• Prepare clean text for AI/LLM pipelines (plain or json)
• Create timestamped transcripts for compliance or QA reviews
• Batch-export transcripts for research and analytics

Notes

• The Actor deduplicates repeated video IDs in one run.
• Transcript availability depends on whether subtitles exist for a given video.
• Leave language empty for the simplest and most reliable result. The Actor will choose a manually created transcript when available, otherwise an auto-generated transcript.
• For best throughput on larger batches, increase maxConcurrency gradually.