·8 min read·GUIDE

How to Download a YouTube Transcript (2026): 5 Free Methods That Work

Five free ways to download or extract a transcript from any YouTube video in 2026. Step-by-step methods that work on desktop, mobile, and for videos without auto-captions. Includes a batch workflow for downloading transcripts at scale.

Michael LiuMichael Liu·
youtube transcriptdownload youtube transcriptyoutube to transcripttranscribe youtubeyoutube to textyt transcript

Searches for "YouTube transcript" climbed to 110,000 a month in the U.S. (+49% year-over-year) and an extraordinary 550,000 a month in India (+307% YoY) — making "how to download a YouTube transcript" one of the highest-volume queries in 2026's video-tooling space. The audience is mostly students, researchers, content creators, and ESL learners who need the text of a video they can't or don't want to re-watch.

YouTube itself has buried the transcript feature behind two clicks and made it copy-resistant. Below are five free methods that work today (May 2026) on desktop, mobile, and for videos where YouTube hasn't generated auto-captions at all.

What "downloading a YouTube transcript" actually means#

A YouTube transcript is the timestamped text of every word spoken in the video. Two flavors:

  1. YouTube's auto-generated captions — produced by Google's speech recognition system, available within minutes of upload, 90-95% accurate. Available in 13+ languages with auto-translate to ~130.
  2. Creator-uploaded transcripts — when the creator uploads their own SRT or VTT subtitle file. More accurate (often near 100%), available immediately, but only present on ~20% of videos.

When most people say "I want to download a YouTube transcript", they mean either: extract the plain text without timestamps (for reading, summarizing, search), or download the timestamped SRT/VTT (for editing into another video, accessibility, or translation work).

The five methods below cover both.

Method 1: YouTube's built-in transcript panel (works for ~80% of videos)#

The official path. Two clicks, plain text, no third-party tool. Best for one-off use on desktop.

  1. Open the YouTube video in any modern browser.
  2. Below the video, click ... moreShow transcript. (On the new YouTube UI as of 2026, this is in the description panel, right under the video title.)
  3. The transcript panel opens on the right side, with each line timestamped.
  4. To get plain text, click the three-dot menu in the transcript panel → Toggle timestamps off → click the first line, scroll to the last line while holding Shift to select all → Cmd+C / Ctrl+C to copy.
  5. Paste into your destination document.

Limitations: the "Show transcript" option doesn't appear if the creator disabled captions, if the video has no spoken content, or if YouTube hasn't generated auto-captions yet (typically the first 30 minutes after upload).

Method 2: youtube-transcript-api (Python, for batches and scripting)#

The fastest method when you need transcripts from multiple videos at once. Free, no signup, scriptable.

pip install youtube-transcript-api
from youtube_transcript_api import YouTubeTranscriptApi

video_id = 'p3q5zWCw8J4'  # the part after v= in the URL
transcript = YouTubeTranscriptApi.get_transcript(video_id)

# Plain text
text = ' '.join([line['text'] for line in transcript])
print(text)

For batch download of, say, an entire channel's back catalog:

import json
from youtube_transcript_api import YouTubeTranscriptApi

video_ids = ['ID1', 'ID2', 'ID3']  # from yt-dlp or YouTube Data API
for vid in video_ids:
    try:
        t = YouTubeTranscriptApi.get_transcript(vid)
        with open(f'{vid}.json', 'w') as f:
            json.dump(t, f)
    except Exception as e:
        print(f'{vid}: {e}')

A 50-video channel takes ~2 minutes to fully download. Works in any country, including India, and against any video that has YouTube captions.

Caveat: YouTube sometimes rate-limits unauthenticated requests from data-center IPs. If you're running this from a cloud VM and hitting 429 errors, run from a residential connection instead.

Method 3: The yt-dlp command line (any captions language, SRT or VTT format)#

For when you want timestamped SRT/VTT in a specific language, including auto-translated ones.

brew install yt-dlp  # macOS, or 'pip install yt-dlp'

# English captions, VTT format:
yt-dlp --write-auto-sub --sub-lang en --skip-download --sub-format vtt \
  "https://www.youtube.com/watch?v=VIDEO_ID"

# Auto-translated Hindi:
yt-dlp --write-auto-sub --sub-lang hi --skip-download --sub-format vtt \
  "https://www.youtube.com/watch?v=VIDEO_ID"

# Plain text (no timestamps):
yt-dlp --write-auto-sub --sub-lang en --skip-download --sub-format vtt \
  "https://www.youtube.com/watch?v=VIDEO_ID"
# then strip the VTT formatting:
grep -v '^[0-9]\+:' VIDEO_ID*.vtt | grep -v WEBVTT | grep -v '^$' | sort -u

yt-dlp is the most powerful free option but requires command-line comfort. The advantage is full control: choose language, format, timestamp behavior, and batch any number of URLs from a text file with --batch-file urls.txt.

Method 4: Voqusa (paste URL, no signup, transcribe even if YouTube has no captions)#

For videos that don't have YouTube captions — a 100% original podcast upload, a music video, a video the creator disabled captions on — methods 1-3 fail. The fallback is to re-transcribe the audio from scratch using a transcription service.

  1. Open Voqusa.
  2. Paste the YouTube URL into the input box.
  3. Click Transcribe. No signup required.
  4. Wait 1-3 minutes for the audio extraction + transcription.
  5. Download as plain text, SRT, or VTT.

Voqusa runs the audio through gpt-4o-transcribe (the same model behind OpenAI's transcription endpoint), so accuracy on clean YouTube audio is 95%+ — often better than YouTube's own auto-captions, which were trained on a noisier corpus. For the full accuracy comparison and Word Error Rate numbers, see our Voqusa vs Otter benchmark.

The URL-paste flow also works for TikTok, Instagram Reels, and Facebook video URLs — useful when you're researching cross-platform content.

Method 5: Browser extensions (one-click on desktop)#

For non-developers who want the easiest possible flow:

  • YouTube Summary with ChatGPT & Claude (Chrome / Edge) — adds a panel to every YouTube video with the transcript and a one-click summary. Free with optional paid tier.
  • YouTube Transcript & Summary (Chrome) — open-source, no account, downloads the transcript as plain text.
  • Glasp (Chrome / Firefox) — highlights and exports YouTube transcripts to a personal knowledge base. Free.

Trade-offs: extensions require granting access to YouTube pages. Read the permissions before installing. The two we trust most for privacy are the open-source "YouTube Transcript & Summary" and Glasp.

How to download a YouTube transcript on mobile#

Mobile is the harder case. YouTube's iOS and Android apps both show the transcript (open a video → tap the description → scroll to the bottom → "Transcript" button) but make copying difficult.

Workarounds:

  • Mobile Chrome → Desktop mode. Tap the menu (⋮ on Android, share icon on iOS) → "Desktop site" → reload. The full desktop UI loads, including the standard transcript panel from Method 1.
  • Share to a transcript tool. Many transcription apps register a share-extension. From the YouTube app, tap Share → choose Voqusa or another transcript tool → paste-URL transcription runs on the URL.
  • YouTube to MP3 then transcribe. As a last resort, download the audio via a YouTube downloader app, then upload to a transcription tool (see our voice recording transcription guide).

How to choose the right method for your use case#

Five common scenarios and the method we'd reach for first:

  1. One-off, casual reading. Method 1 (YouTube's built-in panel). Two clicks, no tools, no friction.
  2. Batch download from an entire channel. Method 2 (Python youtube-transcript-api). 100+ videos in minutes.
  3. Need SRT/VTT in a specific language for video editing. Method 3 (yt-dlp).
  4. Video has no captions, or you need higher accuracy than YouTube's auto-captions. Method 4 (Voqusa or another paste-URL transcription tool).
  5. Non-technical, daily use. Method 5 (browser extension).

A worked example: extracting 50 podcast transcripts for research#

A common research scenario — gather transcripts from a 50-episode podcast back-catalog, search for a specific phrase across all episodes, and export results to a CSV.

  1. Get the video IDs via the YouTube Data API or yt-dlp --get-id PLAYLIST_URL.
  2. Batch-download transcripts with youtube-transcript-api (Method 2). 2 minutes for 50 episodes.
  3. Concatenate and search with a one-line shell command:
    grep -i 'pricing strategy' *.json | jq -r '.text'
    
  4. Export to CSV via Python pandas if needed.

End-to-end: ~15 minutes including writing the search script. Manual listening for the same query would take 50+ hours.

Common problems and fixes#

"Show transcript" doesn't appear in the YouTube UI. Either the creator disabled captions (rare), the video is too new (wait 15-30 minutes), or you're in an older YouTube interface (the transcript button is below the video description; if not visible, click "...more" first).

youtube-transcript-api returns "No transcripts available". The video has no captions and no auto-captions generated. Fall back to Method 4 (paste-URL transcription).

yt-dlp says "Sign in to confirm you're not a bot". YouTube is rate-limiting your IP. Solutions: (1) wait 24 hours; (2) run from a different network; (3) authenticate yt-dlp with browser cookies (--cookies-from-browser chrome).

The transcript is in the wrong language. YouTube auto-translates captions when you select a non-source language. If accuracy is critical, get the source-language transcript and translate it yourself with a dedicated tool — see our AI audio translation guide for the cross-lingual workflow.

Frequently asked questions#

Is downloading a YouTube transcript free? All five methods above are free. YouTube itself doesn't charge for the built-in transcript panel; youtube-transcript-api and yt-dlp are free and open-source; Voqusa's URL-paste transcription is free with no signup; the browser extensions are free with optional paid tiers.

Is it legal to download a YouTube transcript? For personal use (study, research, accessibility), yes — captions are typically considered part of the publicly-available video. For commercial use or republishing, copyright law applies to the spoken content itself. Quote-and-attribute works in journalism and research; wholesale republishing of transcripts does not.

Can I download a YouTube transcript in Hindi, Tamil, or Bengali? Yes. YouTube auto-generates captions in 13+ source languages and auto-translates to ~130. Use Method 2 or 3 with the appropriate language code (hi, ta, bn). For higher accuracy than YouTube's auto-captions on Indian languages, re-transcribe via Method 4.

Why doesn't the YouTube transcript copy correctly? YouTube intentionally makes the transcript panel difficult to copy in one action — they don't want bulk export of captions for SEO or competitor analysis. Workaround: click the first line, scroll to the last, Shift+click the last line, then Cmd/Ctrl+C. Or use any of the four other methods, which all output clean plain text.

Can I download a transcript without timestamps? Yes. In Method 1, toggle timestamps off in the panel menu. In Method 2 (Python), join the text fields without start/duration. In Method 3 (yt-dlp), post-process the VTT to strip timestamp lines. The plain-text version is what you want for reading, summarizing, and most research use cases.

How accurate are YouTube's auto-captions? On clean English audio, YouTube auto-captions hit 90-95% Word Error Rate. Accuracy drops 5-15 percentage points on accented speech, noisy environments, or non-English audio. For higher accuracy, use a dedicated transcription service — see Voqusa vs Otter benchmark for the WER comparison on a controlled test.

Where to start#

For most "I just want to read what this video said" cases, Method 1 (YouTube's built-in panel) is enough. Two clicks, no tools, plain text.

For research, batch work, or higher accuracy than YouTube provides, escalate to Method 2 (Python), Method 3 (yt-dlp), or Method 4 (paste-URL transcription via Voqusa — works on the same URL, no signup, audio re-transcribed at higher accuracy if needed).

The 550,000-monthly-searches volume in India and 110,000 in the U.S. tell you this is a need that isn't going away. Bookmark the method that fits your workflow and stop re-watching videos to find the one line you needed.