YouTube Captions by the Numbers: What 30,000 Transcripts Show (2026)
How many YouTube captions are auto-generated, how fast people talk (words per minute by language) and how much is said in a video of a minute or less, counted from nearly 30,000 videos. Sample and method stated, with lines you can cite.
00:04:00 · NOV 01, 2026
The short version
- Of 24,478 videos with a known caption type, 94.9% had auto-generated captions and 5.1% had captions the creator wrote.
- The median English video is spoken at 163 words per minute, so 10 minutes is about 1,630 words.
- A video of one minute or less holds a median of 91 English words.
What we counted
GetYouTubeTranscript stores the transcript of every video that goes through it. Between 1 May 2025 and 1 November 2026 that came to 29,974 videos from 6,605 channels. For each one we kept the caption type (auto-generated or written by the creator), the caption language, the length and the word count, and we counted.
Pace figures use the 23,009 videos longer than one minute, from 4,629 channels, as words divided by minutes. The short-video figures use the 3,113 videos of one minute or less. We cannot tell a Short from any other short video, and Shorts can now run up to three minutes, so this is a stand-in. The median video over a minute is about 19 minutes long.
Almost every caption is auto-generated
24,478 of the videos have a recorded caption type. Of those, 23,238 (94.9%) have auto-generated captions and 1,240 (5.1%) have captions the creator wrote or uploaded.
That matters if you build on transcripts. Auto-generated captions have no punctuation guarantees and can misspell names and product terms, so a search over transcripts or a summary built on them should expect that. A video with no captions at all returns nothing to read, and the API reports that as an error instead of guessing.
How fast people talk on YouTube
Median speaking pace in captions, videos longer than one minute. English runs at 163 words per minute; auto-generated English captions come in at 162 and creator-written ones at 153.
| Caption language | Videos | Median words per minute |
|---|---|---|
| English | 19,113 | 163 |
| Portuguese | 890 | 146 |
| Russian | 317 | 152 |
| Spanish | 266 | 149 |
Languages with fewer than 250 videos are left out. Japanese is left out because its words are not separated by spaces, so a word count is not comparable.
How much is said in a video of a minute or less
An English video of one minute or less has a median of 91 words, spoken at 153 words per minute, a little slower than long videos. That is about one paragraph. Most YouTube Shorts fall in this range. To get the text of one, paste its link into the YouTube Shorts transcript tool.
Cite these numbers
Each line below stands on its own, with the sample named, so you can quote it as written.
- In 24,478 YouTube videos fetched through GetYouTubeTranscript between May 2025 and November 2026, 94.9% had auto-generated captions and 5.1% had creator-written captions.
- The median English YouTube video longer than one minute is spoken at 163 words per minute (GetYouTubeTranscript, 19,113 videos).
- The median English YouTube video of one minute or less holds 91 words (GetYouTubeTranscript, videos of one minute or less, a stand-in for Shorts).
This page keeps the same URL when the numbers are updated, so a link to it stays valid.
What these numbers do not show
- The sample is the videos people fetched through our service, not a random draw from all of YouTube. Channels people ask for transcripts of lean toward interviews, tutorials and talks.
- Caption type is recorded for 24,478 of 29,974 videos; older rows from before we recorded it are left out of the auto versus creator-written split.
- Words per minute is the caption word count divided by the video length, so silence and music lower it.
- No data about viewers, accounts or who requested a video is used.
Questions about this data
How many YouTube videos have auto-generated captions?
In our sample of 24,478 videos with a known caption type, 94.9% had auto-generated captions and 5.1% had captions written by the creator. The sample is the videos people fetched through GetYouTubeTranscript, not a random draw from all of YouTube.
How many words per minute do YouTubers speak?
The median English video over one minute runs at 163 words per minute in its captions (162 for auto-generated captions, 153 for creator-written ones). A 20-minute video is therefore about 3,200 words.
How long is a typical YouTube transcript?
The median video over one minute in our sample is about 19 minutes long, and its English transcript has a median of 3,123 words. A video of one minute or less (where most Shorts fall) has a median of 91 English words.
Where does this data come from?
From the transcripts stored by GetYouTubeTranscript between 1 May 2025 and 1 November 2026: 29,974 videos from 6,605 channels. We counted caption type, language, length and word count. Nothing about viewers or accounts is included.
Related
- YouTube "Show Transcript" Button Missing? Where It Is Now and Why It Disappears
- How to Cite a YouTube Video in APA, MLA and Chicago (With Timestamps and Transcript Quotes)
- Faceless YouTube Channel with AI: Research with Transcripts, Write Scripts YouTube Will Pay For
- How to Translate a YouTube Video: Transcript and Subtitles in Any Language
- Can ChatGPT Watch YouTube Videos? How to Connect YouTube to ChatGPT (2026)
- How to Get a YouTube Transcript for Claude (2026): Can Claude Watch YouTube Videos?
- We Benchmarked Three Ways to Get a YouTube Transcript: Home Connection vs Cloud Server
- youtube-transcript-api Errors Explained: IpBlocked, RequestBlocked and More
- 191 Starter Story Founder Interviews: How They Got Their First Customers
- How to Get a YouTube Video Transcript: Find, View, Copy and Download (2026)
- n8n YouTube Transcript Failing on Your Server? Why YouTube Blocks It and How to Fix It
- YouTube API Quota Exceeded: Causes and Fixes
- YouTube Transcript API Rate Limit: What It Is and How to Handle 429s
- YouTube Transcript MCP Server: Setup Guide for Claude and Other AI Tools
- YouTube Transcripts in n8n: HTTP Request Workflow Guide
- Migrating from Supadata: A Field Guide
- How to Get YouTube Transcripts in Python (youtube-transcript-api and a REST API)
- Migrate from TranscriptAPI.com: Parameter Mapping and curl Example
- Build an AI YouTube Video Summarizer with LangChain and Next.js
- How to Build a RAG Pipeline with YouTube Transcripts
- YouTube Transcript API: 6 Options Compared (2026), Free and Paid
- yt-dlp Subtitles: Download YouTube Subtitles and Transcripts (Tested Commands, 2026)
- YouTube Automatic Captions: Turn Them On or Off, Fix Missing Ones, Get the Text (2026)
- Can Gemini Watch YouTube Videos? What the App and the API Actually Do (2026)
- Can Copilot Watch YouTube Videos? Edge Sidebar vs the Copilot App (2026)
- Summarize.tech Not Working? It Has Been Down Since August 2026 (Alternatives)
- Can Grok Watch YouTube Videos? Give Grok YouTube Transcripts with MCP (2026)
- Can Perplexity Watch YouTube Videos? Add a YouTube Transcript Connector (2026)
- GetYouTubeTranscript vs TranscriptAPI
- GetYouTubeTranscript vs youtubetotranscript.com
- GetYouTubeTranscript vs youtube-transcript.io
- GetYouTubeTranscript vs NoteGPT
- GetYouTubeTranscript vs Tactiq
- GetYouTubeTranscript vs Glasp
- GetYouTubeTranscript vs Eightify
- GetYouTubeTranscript vs Supadata
- GetYouTubeTranscript vs DownSub
- GetYouTubeTranscript vs youtubetranscript.dev

































































