TikTok Transcript Scraper: the spoken words from a TikTok video, with timings, SRT and VTT

Paste TikTok links and get the words back as text: the full transcript, the same text as timed segments, and ready-made SRT and WebVTT files. Each row also carries the video id, author, caption, hashtags, duration, play count and the language the track is in.

The thing to know before you plan a run: this reads the caption track TikTok already holds for a video. It does not listen to audio. A video with no caption track has nothing to give, and you get an uncharged row saying so. In a 120-video sample pulled from TikTok's own topic pages, 85 had a caption track and 35 did not.

Input	Video links, vm./vt. share links, or bare video ids
Output	One row per transcript, with segments, SRT and WebVTT
Ceiling	1,000 links in, 5,000 transcripts per run
Account needed	None for almost every video
Price	$0.40 per 1,000 transcripts, flat on every plan

🗣️ What TikTok Transcript Scraper does

It opens each video's own page, reads the caption track out of it, and turns that into four shapes of the same thing: one block of text, an array of segments with start and end times, an srt string and a vtt string. Drop the SRT straight into an editor, or work off the segments.

TikTok makes a caption track itself when it hears speech, and creators can upload their own. Where TikTok also holds a translated track you can ask for it with language, and the row comes back marked isTranslated: true. When there is no translation you get the original rather than an error.

📥 What you give it

{
  "videoUrls": [
    "https://www.tiktok.com/@studywithlizzz/video/7657803613355412766",
    "7516208601434819862"
  ],
  "maxItems": 10,
  "language": ""
}

Field	Default	What it is
videoUrls	box holds two example links	Full links, vm./vt. share links or bare numeric ids, mixed freely. Up to 1,000 per run.
maxItems	box holds 10	How many transcripts to return, up to 5,000. The list is read in order and the run stops when it has enough, so links past that point cost nothing.
language	empty	A code like es, fr, de or ja. You get the translated track where TikTok has one, the original where it does not.
concurrency	6, up to 12	How many videos are read at once. Lower it for a very long list.
sessionCookies	empty	Optional. Your own TikTok sessionid, for the few videos shown only to a signed-in visitor. Anyone holding that value can act as your account, so treat it like a password.
proxyUrls	empty	Optional. Your own servers, as http://user:pass@host:port, used exactly as given.

Run it with the input empty and you get one labelled sample row and nothing else.

📤 What you get back

A real row from a recent run. The long fields and one name carrying emoji are cut short here:

{
  "ok": true,
  "charged": true,
  "recordType": "transcript",
  "videoId": "7657803613355412766",
  "url": "https://www.tiktok.com/@studywithlizzz/video/7657803613355412766",
  "authorUsername": "studywithlizzz",
  "authorName": "liz | studytok ...",
  "caption": "sometimes studying longer doesnt always mean higher grades...  #studytok #studytips",
  "hashtags": ["studytok", "studytips", "studyhacks", "studymotivation", "examtips"],
  "durationSeconds": 63,
  "createdAt": "2026-07-02T05:49:24.000Z",
  "language": "eng-US",
  "languageCode": "en",
  "languageName": "English",
  "isAutoGenerated": true,
  "isTranslated": false,
  "captionSource": "ASR",
  "text": "You scored at the top of your class because you studied for three hours the night before your exam, or so you thought...",
  "wordCount": 238,
  "characterCount": 1292,
  "segmentCount": 31,
  "segments": [{"start": 0, "end": 1.58, "startTime": "00:00:00.000", "endTime": "00:00:01.580", "text": "You scored at the top of your class"}],
  "srt": "1\n00:00:00,000 --> 00:00:01,580\nYou scored at the top of your class\n\n...",
  "vtt": "WEBVTT\n\n00:00:00.000 --> 00:00:01.580\nYou scored at the top of your class\n\n...",
  "availableLanguages": ["eng-US"],
  "playCount": 3600000,
  "likeCount": 455300,
  "commentCount": 1285,
  "shareCount": 21000,
  "musicTitle": "original sound",
  "scrapedAt": "2026-09-21T01:29:19.131Z"
}

Field	What it is
text	Every cue joined into one block, which is what you want for search or a model.
segments	Cue-level timings. start and end are seconds, startTime and endTime the timecode form.
srt, vtt	The same cues as subtitle files, ready to save.
isAutoGenerated, captionSource	Whether TikTok's own speech recognition made the track, or a person uploaded it.
availableLanguages	Every caption track that video exposed on that request, so you can see what else you could ask for.
playCount, likeCount, commentCount, shareCount	Counts at read time. A genuine zero comes back as null.
inputUrl, url	What you pasted, and the canonical link with the real handle.
coverUrl	The video's cover image. The link carries an expiry.

🧾 Reading the output

Three kinds of row can land in your dataset.

Row	How to spot it
A transcript	ok: true and recordType: "transcript"
The sample row	_sample: true, written only when the input was empty
A diagnostic	ok: false, _diagnostic: true and an errorCode

charged: true marks a real transcript row and false marks a sample or diagnostic one. Read it as the row type: it is stamped as the row is built, so it is not a receipt. Your run's own event count in Apify is the billing record.

Code	What it means
NO_TRANSCRIPT	The video has no caption track. Photo slideshows land here too.
NOT_FOUND	Deleted, private, region-locked, or shown only to a signed-in visitor.
BAD_INPUT	That line is not a TikTok video link or id.
BLOCKED	TikTok would not serve that page this time. Worth a re-run.
NETWORK	The page was unreachable or answered badly. A share link whose redirect fails lands here.
NO_RESULTS	Too many videos in a row had no caption track, so the run stopped early instead of working through the list.
TIME_BUDGET	The run ran out of time before reaching that video.
PROXY_INPUT_ADJUSTED	Something you put in proxyUrls was unusable, so the run carried on without it.
CHARGE_ERROR	A charge could not be recorded. A couple in a row stops the run.

▶️ How to run it

1. Open TikTok Transcript Scraper and click Try for free.
2. Paste your links into TikTok videos, one per line, replacing the two examples.
3. Leave Preferred language empty for the original words, or put a code like es in it.
4. Set Maximum transcripts low for the first run, then click Start.
5. Download the dataset as JSON, CSV or Excel, or read it from the Apify API.

💰 How much does it cost?

$0.40 per 1,000 transcripts. Flat on every Apify plan, no volume tiers.

You are charged per transcript delivered. A video with no caption track, a deleted or private one, a line that was not a link, and the sample row are all uncharged. Links past your maxItems are never opened.

One thing to watch: the same video pasted twice, once as a share link and once as a full link, is read twice and counts twice. Paste one form per video.

💡 What people use it for

• Turning a set of videos into text you can search, quote or feed to a model.
• Making subtitle files for your own uploads, straight from srt or vtt.
• Finding which videos in a niche actually say a phrase, rather than just tagging it.
• Pulling the hook out of high-performing videos: the first few segments are the first few seconds.
• Reading what a competitor says in their videos without watching all of them.

🚧 What it does not do

• It does not listen to audio. No caption track means no transcript, and about one video in three has none.
• Timings are cue-level, not word-level. Each segment covers a phrase.
• On-screen text is not included. Stickers, overlays and burned-in captions are pictures, not a caption track.
• Translations only exist where TikTok made one. Asking for a language it does not hold gives you the original track, marked isTranslated: false.
• No live videos or stories.
• Age-restricted and private videos need your own cookie, and come back as uncharged NOT_FOUND rows without one.
• A zero count reads as null. A video with genuinely no comments shows commentCount: null.
• If every line you paste is unusable, you get the sample row rather than a message per line. Check your list if a run comes back with one row.
• Counts are a snapshot, and a popular video's play count moves while you read it.

🧭 Which TikTok tool do you need?

If you want	Use
The spoken words from a video	This one
The video file itself, as MP4 or MP3	TikTok Video Downloader
Comments under a video	TikTok Comments Scraper
Videos under a hashtag, to collect links first	TikTok Hashtag Scraper
The same job on YouTube	YouTube Transcript Scraper

❓ Questions people ask

Do I need a TikTok account or an API key? No. Almost every video is read with nothing signed in. The cookie field is only for the few that are gated.

Why did some videos come back empty? They have no caption track. TikTok makes one when it hears speech, so silent, music-only and some slideshow posts have nothing to read.

Can I get a Spanish transcript of an English video? Only if TikTok already holds a Spanish track for it. Nothing is translated here.

Can I feed it a whole profile or hashtag? Not directly. Collect the video links first, then paste them in.

Can I run it on a schedule? Yes. videoId is stable, so a transcript you already have is easy to skip.

Is this legal? The captions are published with the public video. Copyright in the words belongs to whoever said them, so quoting and analysis are safer ground than republishing. Apify's write-up on scraping and the law is a good starting point, and we are not lawyers.

🆘 If something breaks

Open the Issues tab on the actor page. Send the video link and the run ID. The errorCode on the diagnostic row usually names the problem on its own.