YouTube Transcript Search — Caption Keyword Finder & Clip Extractor

Search YouTube video transcripts and captions for any keyword or phrase. Scan entire channels to find exactly when a topic is mentioned — with exact timestamps, transcript context, and clip download commands.

Stop scrubbing through hours of video. This YouTube transcript search tool scans every video in a channel and tells you the exact second a word was spoken — across hundreds of videos in minutes.

🎯 Best For

• Content research — find every video that mentions your topic, with exact timecodes
• Brand monitoring — track brand mentions across YouTube channels
• Competitive analysis — see what competitors say about specific subjects
• Academic research — locate references to papers, theories, or people across lecture channels
• Clip extraction — get download commands for specific video segments
• Podcast production — find sound bites and quotes from interview channels

📊 What You Get

• Exact timestamps where the keyword appears in the transcript
• Clickable YouTube links that jump to that moment in the video
• Transcript excerpts with keywords highlighted
• yt-dlp download commands for each matching clip
• Interactive HTML report with video thumbnails and copy buttons
• CSV export for spreadsheets and data analysis
• Batch download scripts (.bat and .sh) to grab all clips at once

⚙️ How It Works

Step 1 — Enter channels and keywords

Paste YouTube channel URLs and the words or phrases to search for. Supports multiple channels and multiple keywords — comma-separated or one per line.

Step 2 — Automatic transcript scanning

The Actor fetches every video's transcript and searches for your keywords. YouTube Shorts are filtered out automatically. Results stream to your dataset in real time.

Step 3 — Browse and download

Open the HTML report to preview results visually. Click thumbnails to preview on YouTube. Copy yt-dlp commands to download clips. Export as CSV.

🔑 Key Features

• YouTube transcript search — full-text search across all video captions
• YouTube caption search — works with auto-generated and manual captions
• Multiple keywords — search for several terms at once
• Multiple channels — scan across several channels in one run
• Smart clip merging — nearby mentions combine into single clips
• Minimum clip duration — filter out brief passing mentions
• Exclude Shorts — skip videos under 60 seconds
• HTML report — interactive report with thumbnails and copy buttons
• CSV export — spreadsheet-ready results
• Batch download — scripts to download all clips at once
• Abort-safe — results saved every 25 videos
• No API key — works without YouTube Data API credentials
• Residential proxies — avoids YouTube rate limiting

📝 Input

Parameter	Type	Required	Default	Description
channelUrls	string	Yes	—	YouTube channel URLs, comma-separated or one per line
keywords	string	Yes	—	Search terms, comma-separated
maxVideos	integer	No	all	Max videos to scan per channel
excludeShorts	boolean	No	true	Skip YouTube Shorts
minClipDurationSec	integer	No	0	Minimum clip length in seconds
snippetPaddingSec	integer	No	15	Padding before/after match
mergeGapSec	integer	No	30	Merge mentions within this gap
contextLines	integer	No	3	Context lines around each match
language	string	No	en	Transcript language preference

Example Input

{
    "channelUrls": "https://www.youtube.com/@veritasium",
    "keywords": "quantum, relativity",
    "maxVideos": 100
}

💰 Pricing

Apify charges for compute time, proxy bandwidth, and storage. Typical costs:

Videos scanned	Approximate cost
10 videos	~$0.05
50 videos	~$0.25
100 videos	~$0.50
500 videos	~$2.50

Costs vary based on transcript length, proxy usage, and result count. Set a max cost per run in Apify to control spending.

📤 Sample Output

{
    "channelName": "veritasium",
    "videoTitle": "The Internet Was Weeks Away From Disaster and No One Knew",
    "keyword": "science",
    "timestampFormatted": "00:05:12",
    "videoUrlAtTimestamp": "https://www.youtube.com/watch?v=aoag03mSuXQ&t=312",
    "transcriptExcerpt": "...was a young computer **science** student who just happened to be building his own kernel...",
    "mentionCount": 1,
    "ytDlpCommand": "yt-dlp --download-sections \"*00:04:57-00:05:30\" -o \"The_Internet_Was_Weeks_Away_00-04-57.%(ext)s\" \"https://www.youtube.com/watch?v=aoag03mSuXQ\""
}

🏢 Built for Teams & Automation

• API integration — trigger runs via the Apify API
• Scheduling — recurring searches on any cadence
• Webhooks — notifications when runs complete
• Integrations — Google Sheets, Slack, Zapier, Make, and more

❓ FAQ

Does it work with auto-generated captions? Yes — uses whatever transcript is available, including YouTube's auto-generated captions.

What languages are supported? Any language with available transcripts. Set the language parameter.

How long does a full channel take? About 15-25 minutes for 500 videos. Use maxVideos for faster runs.

What if I stop the run early? Results save every 25 videos — you keep everything including the HTML report and CSV.

Can I search multiple channels? Yes — comma-separate or one per line.

Do I need a YouTube API key? No. No Google API credentials required.

Why are some videos skipped? Videos without transcripts, age-restricted videos, or temporary YouTube rate limits. The Actor logs which videos are skipped and why.

🔄 How Does This Compare?

vs YouTube's built-in search — YouTube finds videos about a topic. This Actor finds the exact second a word is spoken inside a video.

vs full transcript scrapers — Most scrapers dump raw text. This Actor searches it and returns only the relevant moments with timestamps and download commands.

vs manual searching — Opening each video and skimming captions takes hours. This scans an entire channel in minutes.

🔧 Technical Details

• scrapetube for channel video listing (no API key)
• youtube-transcript-api for timestamped transcripts
• Residential proxy rotation per request
• 2-minute timeout on channel listing
• Auto-saves every 25 videos
• 1.5s delay between requests