Video and content research
Turn videos into transcripts and on-screen text you can search and summarize.
Pass a video URL and get back what is said in it, line by line with timestamps. Add the text written on screen where the words matter more than the voice, then search or summarize the result.
Start with a template
Transcribe a video
The transcript of one video as a ready program in TypeScript and Python.
Video transcription cookbook
One function that returns the spoken words of a video from any supported platform.
Run your first request
This reads the transcript of a public YouTube video.
curl "https://www.socialcrawl.dev/v1/youtube/video/transcript?url=https://www.youtube.com/watch?v=dQw4w9WgXcQ" \
-H "x-api-key: sc_your_key_here"import requests
response = requests.get(
"https://www.socialcrawl.dev/v1/youtube/video/transcript",
params={"url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ"},
headers={"x-api-key": "sc_your_key_here"},
)
for line in response.json()["data"]["transcript"]:
print(line["offset"], line["text"])const url = new URL("https://www.socialcrawl.dev/v1/youtube/video/transcript");
url.searchParams.set("url", "https://www.youtube.com/watch?v=dQw4w9WgXcQ");
const response = await fetch(url, { headers: { "x-api-key": "sc_your_key_here" } });
const { data } = await response.json();
for (const line of data.transcript) {
console.log(line.offset, line.text);
}Each line carries its text, the offset in seconds where it starts and its duration. A transcript is priced per video, not per minute. This call costs 3 credits, whatever the length. To see the price first, pass the same url to GET /v1/utility/estimate with id=youtube/video/transcript, which is free. Endpoint pricing lists every call.
To see which caption languages a YouTube video has before you read one, call /v1/youtube/video/subtitles. It lists each caption track with its language and a download link. A transcript you have read before is usually served from cache, and a cache hit is free. Caching says how long each kind of data stays fresh.
How it works
- Pass a video URL to the transcript endpoint of its platform. TikTok, Instagram, YouTube, Facebook, X, Reddit, LinkedIn and Rumble each have one.
- Read the text with its timestamps. On TikTok,
use_ai_as_fallback=truetranscribes the audio of a video that has no captions. - Add the words shown on screen for TikTok videos and Instagram reels, then search or summarize both together.
Why teams choose SocialCrawl
The price does not grow with the video. Each transcript endpoint charges one flat price per successful call, whatever the length. See transcripts cost the same at any length.
A video with nothing to read costs nothing. A video with no captions or no speech, or one that was deleted, is refunded. See when credits are refunded.
On-screen text has its own endpoints. The words a creator puts on screen never reach the caption or the transcript. The screen text cookbook reads them second by second.
Questions
Can I read the text shown on screen in a video?
Yes, on TikTok videos and Instagram reels. The screen text endpoints return the words written on screen. With frames=all they read the whole video, and each block of text carries the second it first appears. The screen text cookbook runs both and puts the voiceover beside them.
