YouTube transcript: get, copy, and download any video's text

Get a transcript straight from YouTube

YouTube's watch page has a built-in transcript panel. Since the October 2025 layout update it lives inside the expanded description, not under the video: click “…more” under the title, scroll to the bottom of the expanded description, and select Show transcript. The panel opens beside the player with timestamped, clickable lines — click a line and the player jumps to that moment. On the mobile app the transcript replaces the description below the player, with manual selection only.

Copying the whole transcript: click anywhere inside the panel, press Ctrl+A (Cmd+A on Mac), then Ctrl+C. To strip timestamps, use the panel's three-dot menu Toggle timestamps before copying — though in some browsers timestamps come along anyway and need a cleanup pass in your editor.

When the button is missing, the fast diagnosis is the CC button: if captions play on screen a track exists and the control is somewhere in the description; if CC does nothing, the video has no caption track and no extraction tool can conjure one. Brand-new uploads (captions still processing), music videos, and creators who turned captions off are the usual cases — skip to transcribing the audio.

If you own the video, YouTube Studio's captions editor is the official source: it lists auto-generated captions for download as .srt/.vtt, and it is where you upload or fix tracks. Auto-captions generate in the video's default language only.

Transcript generators: what the free tools actually do

Search “youtube transcript” and the SERP is wall-to-wall paste-a-URL generators — youtube-transcript.io (25 free extractions), NoteLM (9 export formats), MeetWave, fbutube, VexaScribe, and dozens of clones. They all do the same job: fetch the caption track YouTube already published and reformat it. That's why the extraction ones are genuinely free — there is no speech recognition to pay for. The fine print lives at the edges:

Two structural facts the tool landing pages don't print. First, extraction inherits YouTube's failures: when the tool says “no transcript available,” that is YouTube answering — the track doesn't exist. Second, every one of these tools processes the video on their servers, which matters when the footage is unreleased, internal, or otherwise not something you'd paste into a public form.

Download formats: TXT, SRT, VTT, JSON — which to pick

The built-in panel gives you copy-paste text only — no file, no format choice. That gap is what the generator tools sell. Each format has a real job:

One nuance worth knowing before you pay for a converter: the SRT/VTT already exist. Every caption track YouTube serves is a timed subtitle file under the hood — the generator is reformatting a track that ships in exactly these formats, which is why conversion is instant and free. For your own videos, YouTube Studio downloads the original caption files directly.

Where formats actually matter for quality: auto-generated tracks carry no punctuation and mishear proper nouns, so an SRT built from the auto-track bakes those errors into your video editor. Creator-uploaded captions are markedly better. If the source track is auto-generated and the text is going anywhere public, expect a cleanup pass.

No captions? Transcribe the audio itself

About one video in ten has no usable caption track — fresh uploads, disabled captions, languages auto-captions doesn't cover. No extractor works there; a speech-recognition engine has to listen to the audio. Two routes:

Route A — a cloud transcriber. Paste the URL into one of the tools above that runs Whisper server-side (or extract the audio and upload it). Fast for a one-off public video; the costs are a meter (free trials sized at 2–3 transcriptions, then per-minute pools) and the upload itself.

Route B — the local pipeline. Download the audio and transcribe on your own machine. Nothing uploads, no meter, no caps:

# 1. pull just the audio, 16 kHz mono WAV (what Whisper-class engines expect)
yt-dlp -x --audio-format wav --postprocessor-args "-ar 16000 -ac 1" \
  -o "audio.%(ext)s" "https://www.youtube.com/watch?v=VIDEO_ID"

# 2. transcribe locally with whisper.cpp (free, CPU or GPU)
whisper-cli -m ggml-large-v3-turbo.bin -f audio.wav -l en -osrt

GPU transcribes a 30-minute video in a couple of minutes; CPU runs ~0.3× real-time. Pick large-v3 for jargon-dense content, small or medium for drafts. This two-command pipeline is the entire product behind the open-source “local caption” wrappers.

Prefer an app to a pipeline? VocalFuse runs a Whisper-class engine locally on Windows: drop in any audio or video file (download the audio with yt-dlp first for YouTube), get a timestamped, punctuated transcript with speaker labels, and export TXT/SRT/VTT. Nothing uploads — there is no per-minute meter because the engine runs on your PC (free tier, Basic $5/mo dictation, Pro $10/mo adds AI notes and summaries).

For developers: the transcript API landscape

There is no official YouTube transcript API — Google has never shipped one. The de-facto standard is the open-source Python library youtube-transcript-api (MIT, v1.2.x), which calls the same caption endpoint the browser uses. Pin 1.2.x and use the fetch()/list() instance methods with cue objects — the pre-1.0 get_transcript() static call still printed in older tutorials raises AttributeError on current installs. Formatters output plain text, JSON, SRT, and VTT.

The production wall is IP-based: extraction works fine from a residential laptop for hundreds of requests, then cloud IPs start hitting RequestBlocked / IpBlocked — AWS Lambda, Cloud Run, and serverless functions almost always fail without residential proxies. That wall is why the hosted-wrapper industry exists. Newer alternatives ride YouTube's internal innertube API instead (the ytranscript CLI also ships an MCP server so coding agents can pull transcripts directly), and yt-dlp can write subtitle files with --write-auto-subs --sub-format srt when you just want the track as a file.

Error set worth handling: TranscriptsDisabled (creator turned captions off), NoTranscriptAvailable, VideoUnavailable (private, deleted, region-locked); age-restricted videos need authentication. Caption extraction rides YouTube's internal timedtext endpoint, which sits in a legal gray zone at volume — architect for breakage.

What the transcript is good for

Quality caution: auto-captions cluster mishears around proper nouns and ship with no punctuation. If the text is going anywhere public — a quote in an article, a name in show notes — verify against the audio before it ships.

Related reading

YouTube transcript — FAQ