/tiktok/video/transcriptGet Video Transcript
Extract the spoken transcript from a TikTok video. Returns timestamped text segments for any video with speech.
Overview
Returns the full TikTok video transcript as raw text plus the source language. The endpoint reads TikTok's own caption layer (the same one the in-app accessibility setting exposes), so you get the creator's actual spoken words rather than a re-transcription of the audio. Average response time is under 2 seconds.
If you've been transcribing TikTok videos manually — copy the link, open a transcript generator, paste, wait, copy back — this is the API behind that workflow. It scales from one video to hundreds of thousands without changing the call shape.
The transcript comes back word-for-word. Filler words, repeated phrases, all of it. We don't summarise, translate, or clean. If you want any of that, post-process the output with your own model — you keep the credit cost predictable and you control the transformation.
Quick Start
curl "https://creatorcrawl.com/api/tiktok/video/transcript?url=https://www.tiktok.com/@creator/video/123" \
-H "x-api-key: YOUR_API_KEY"Parameters
| Name | Type | Required | Description |
|---|---|---|---|
url | string | Required | TikTok video URL |
Response
{
"transcript": [
{
"start": 0,
"end": 2.5,
"text": "Hey everyone, today I want to show you..."
},
{
"start": 2.5,
"end": 5.1,
"text": "how to build an AI agent that uses TikTok data"
}
]
}Every request costs 1 credit. Response data is live -- never cached.
What you can build
Bulk TikTok transcript generator
Wrap the endpoint in a UI and you have the same product as every standalone "TikTok video transcript generator" tool — except yours scales beyond a single URL at a time and can be embedded in a creator workflow or agency dashboard.
Searchable creator archives
Index every transcript a creator has published, then run vector search across the corpus. Lets agencies pitch with "47 mentions of [topic] across our roster" and lets creators surface their own old hooks instantly.
Brand mention monitoring
Combine search-keyword + get-video-transcript to catch every video that says your brand name out loud, including the long tail that never tags your account or uses your hashtag. Sentiment dashboards miss most of these.
Trend research at scale
Pull transcripts for the top videos in a niche, cluster by topic. You see which framings are converting (which hook, which story arc) instead of guessing from view counts.
Training data for short-form models
If you're fine-tuning on short-form voice or hook structure, transcripts are the cheapest source of labelled native-platform language. One credit per video, no scraping.
Use Cases
Content indexing
Make TikTok video content searchable by extracting and indexing the spoken word.
AI training data
Use transcripts as training data for AI models analyzing creator content.
Accessibility
Generate captions or summaries from video transcripts for accessibility purposes.
Code Examples
import requests
resp = requests.get(
"https://creatorcrawl.com/api/tiktok/video/transcript",
params={"url": "https://www.tiktok.com/@creator/video/123"},
headers={"x-api-key": "YOUR_API_KEY"},
)
for segment in resp.json()["transcript"]:
print(f"[{segment['start']:.1f}s] {segment['text']}")const resp = await fetch(
"https://creatorcrawl.com/api/tiktok/video/transcript?url=https://www.tiktok.com/@creator/video/123",
{ headers: { "x-api-key": "YOUR_API_KEY" } }
)
const { transcript } = await resp.json()
transcript.forEach(s => console.log(`[${s.start.toFixed(1)}s] ${s.text}`))Troubleshooting
Empty transcript for a video that has visible on-screen text
On-screen text (the kind creators add via the text overlay tool) is not the same as captions. We return the audio caption layer, not OCR'd overlay text. For overlay text, run a separate OCR pass on the video frames.
Transcript looks truncated mid-sentence
TikTok's caption layer occasionally drops the final 1-2 seconds if the video ends abruptly. This is a TikTok-side limitation. The transcript you receive is exactly what TikTok serves to its own accessibility layer.
404 on a video URL that loads in a browser
Strip everything after the video ID. Canonical form is https://www.tiktok.com/@handle/video/123456789. If it still 404s, the video may be deleted, private, or geo-blocked in our request region.
Frequently asked questions
How do I get a transcript of a TikTok video?
Manually: open the video, hit the share menu, toggle captions on. That works for one video. For anything past 10-20 a week the manual flow falls apart, which is when teams move to the API.
What if the video doesn't have captions enabled?
TikTok auto-generates captions for almost every public video uploaded after late 2022, even if the creator never turned them on. We return those auto-generated transcripts. For old videos or videos where the creator explicitly disabled captions, the response is empty and we do not charge a credit.
Does this transcribe TikTok videos in other languages?
Yes. Transcript comes back in the original language with a language code in the response. Coverage is strong for English, Spanish, Portuguese, French, German, Indonesian, Japanese, Korean, Mandarin, and Arabic. For translation, run the result through any translation API.
How does this compare to running Whisper on the video file?
Whisper is excellent but you have to download the video, transcode, run inference, and store output. A typical Whisper pipeline costs $0.005-$0.02 per minute of video plus engineering time. This endpoint costs 1 credit and returns in under 2 seconds because we read TikTok's existing caption layer rather than re-transcribing audio.
Is the transcript word-for-word?
Yes. Raw caption stream. Filler words, ums, repeats — all included. If you want a clean version, post-process it.
How fresh is the transcript data?
Live. We hit TikTok on every request. There's no caching layer between you and TikTok's caption service.