Start free

TED Talk transcription.Speaker labels and SRT — ted.com's built-ins skip both.

Paste a TED Talk URL from ted.com, a TEDx YouTube channel, or TED-Ed. Get a speaker-labeled transcript, SRT and VTT subtitle files, and timestamped chapter markers — none of which TED.com's own transcript page exports.

Drop a file, or pick one

MP3 · WAV · M4A · MP4 · MOV · MKV · OGG · OPUS · FLAC · WEBM — up to 100 MB anonymously

Paste a link, we’ll fetch the audio

YouTube · TikTok · Vimeo · Twitter · SoundCloud · Spotify · 50+ more

Record straight from your browser

Sign up takes 30 seconds — recording opens right after, in the dashboard.

No card required~90s per 60-min fileSRT · VTT · DOCX · TXTFiles auto-deleted in 24h

↓ Watch what comes out

Paste the talk URL. Get captions back.

TED.com has an official interactive transcript — but it doesn't export SRT, doesn't label the moderator vs. the speaker on Q&A talks, and it stops existing the moment you're working with a raw TEDx YouTube upload.

TED.com URLREC 1 speaker + host · 18:42
auto-detected en-USStudio mainstage · lavalier mic
~90s
Transcript · streaming98% accuracy
S1

So the experiment ran for eighteen months across four cohorts — and the result was nothing like what we predicted.

S1

The control group, the one we expected to fail, outperformed every other arm by a factor of three.

S2

Quick follow-up — were the cohorts blinded to which condition they were in?

S1

Fully double-blind. That's what made the result so hard to dismiss.

98% on mainstage studio audioSRT · VTT · DOCX · TXT · JSON

↓ This is the dashboard

This is what loads when the job finishes.

Same layout as the real dashboard — Summary, full Transcript, Speakers tab, Exports. Key points and action items extracted automatically. Auto-tags on every job.

Try it on your own file — it's free

Three real options · honest comparison

TED.com's transcript. YouTube auto-captions. Or us.

TED.com publishes its own transcript for every mainstage talk in 100+ community-translated languages. YouTube auto-captions cover TEDx and TED-Ed. Both are free. Both are missing things you probably came here for.

Option 01

TED.com official transcript

Free, multilingual, interactive. No subtitle file export and no speaker labels.

Coverageted.com mainstage only
Speaker labelsNone — host/Q&A merged
Languages100+ (volunteer translated)
SRT / VTT exportNot available
Chapter markersNo
CostFree
Best forReading along on the TED.com page itself, in a language a TED translator has already covered.
Option 02

Transcription.Solutions

Paste any TED, TEDx, or TED-Ed URL. Server-side audio pull, speaker labels, subtitle files included.

CoverageTED.com · TEDx YT · TED-Ed
Speaker labelsHost + speaker + audience Q
Languages99, auto-detected
SRT / VTT exportBoth, plus DOCX · TXT · JSON
Chapter markersAI summary + chapters (Pro)
Cost · per min$0.03
Best forRe-captioning a TED talk for a course module, blog embed, or accessibility re-upload — anywhere you need a real .srt file.
Option 03

YouTube auto-captions

Free on every TEDx and TED-Ed upload. No punctuation in caption export, no speaker turns.

CoverageYouTube-hosted only
Speaker labelsNone
LanguagesAuto-translate, accuracy varies
SRT / VTT exportVTT scrape only, no punctuation
Chapter markersOnly if uploader added them
CostFree
Best forCasual viewing of a TEDx talk where you just need rough captions on the YouTube player itself.

Pricing and feature flags accurate as of 2026. TED.com translation coverage depends on the volunteer Open Translation Project.

Specific to TED Talks

Three things that bite people transcribing TED talks on generic tools.

TED audio is clean, but the failure modes are specific — applause breaks, cited researcher names, and host-vs-speaker turns on TED Interview format.

What goes wrong

  1. 1Applause and laughter mid-sentence make generic models hallucinate filler words or drop the next clause entirely. TED audiences clap a lot.
  2. 2Cited researcher names and study titles (Kahneman, Dunning-Kruger, Framingham) get spelled phonetically by tools that don't know they're proper nouns.
  3. 3TED Interview and Q&A formats merge the host and the speaker into one block — there's no diarization on TED.com's own transcript at all.

What to flip here

  1. 1Paste the ted.com or YouTube URL directly. We extract audio server-side, no MP4 download, and applause/laughter get tagged as non-speech events instead of becoming garbage words.
  2. 2Drop researcher names, book titles, and field jargon into Custom vocabulary on the job form. Passed as recognizer hints, not hard substitutions.
  3. 3Diarization is on by default. Host, speaker, and audience-mic questioner each get their own label — rename them in one pass after the job completes.

Recommended job settings for TED talks

Paste a TED URL and these flip on automatically. Override per-job from the form.

Input
URL paste · server-side audio pull
Diarization
Acoustic · 1-3 speakers
Language
Auto-detect · 99 supported
Non-speech events
[applause] [laughter] tagged
Summary
Chapter markers + key points (Pro)
Export
SRT · VTT · DOCX · TXT

Accuracy · real-world numbers

98% on mainstage. Q&A and TEDx field recordings degrade predictably.

TED mainstage talks are about as clean as spoken audio gets — lavalier mics, treated venue, single speaker, prepared delivery. That's the ceiling. TEDx local events and audience Q&A are where the numbers slide. Figures below are from real customer TED files.

98%
TED mainstage, native English

Lavalier mic, treated stage, rehearsed delivery. One speaker, no overlap. The cleanest input class we see in production.

96%
TED mainstage, non-native speaker

Same audio quality, accent variation. Proper nouns and technical citations drop a couple points — custom vocab recovers most of it.

92%
TEDx local event, single speaker

Variable venue acoustics, sometimes handheld mic. Occasional applause artifacts. Subtitle timing still frame-accurate.

87%
TED Interview / Q&A with audience mic

Audience questions on a shared roving mic, host overlap, music stings on TED-Ed. Worst case in our TED data.

Common questions

8 things people ask about TED Talk transcription.

01TED.com already has a transcript on every talk. Why pay for this?+
Because the TED.com transcript is read-only on the page — there's no SRT or VTT download button, no speaker labels on TED Interview or Q&A format, and no way to feed it into your own captioning workflow. If all you need is to read along, ted.com is fine. If you need a subtitle file, you need a transcription pass.
02Can I just paste a ted.com URL, or do I need to download the video first?+
Paste the URL directly. We extract the audio server-side — no MP4 download, no browser extension. Same for TEDx YouTube channels and TED-Ed videos. Just drop the link into the job form.
03Does it work for TEDx talks that are only on YouTube?+
Yes. TEDx local events host on the official TEDx Talks YouTube channel and individual chapter channels. Paste any YouTube URL and we resolve it the same way. Accuracy is a touch lower than mainstage because venue audio varies, but it lands around 92% on most TEDx uploads.
04What about TED-Ed animations with a narrator and background music?+
Works, but it's the hardest case in our TED data. The narrator is usually clean, but music beds and sound design eat a few accuracy points. We tag music as a non-speech event so it doesn't pollute the transcript. Expect around 87-90% on TED-Ed.
05Can I get chapter markers like the ones TED sometimes shows?+
Yes, on Pro and Business plans. The AI summary includes timestamped chapter markers — typically 4 to 8 per talk, depending on length. They're generated from the transcript content, not pulled from TED's own metadata.
06How does it handle applause, laughter, and standing ovations?+
They're tagged as non-speech events — [applause], [laughter], [music] — in the transcript and skipped in subtitle timing. Generic tools without this often hallucinate filler words during applause, which then end up baked into your .srt file.
07Does it work for TED talks in languages other than English?+
Yes — auto-detect handles 99 languages. TED.com's volunteer translations cover more languages than we transcribe in (their Open Translation Project has 100+), but those are translations of the English source. We transcribe the original audio in whatever language the speaker actually spoke.
08Am I allowed to transcribe TED talks under their license?+
TED Talks are published under Creative Commons BY-NC-ND, which permits non-commercial re-use with attribution but not derivative works for commercial distribution. Transcribing for personal study, research, or internal accessibility is generally fine. Re-publishing a transcript or captioned video commercially is not. We can't give you legal advice — check the TED Terms of Use and CC license for your specific use.

Paste a TED Talk URL. See what comes out.

30 free minutes every month. No card. Speaker labels, SRT and VTT exports, chapter markers on Pro.

Start free