TikTok Data Scraper — Search, Comments, Music & Captions

By Influship. Discover creators and videos, monitor audience conversations, and retrieve spoken text for research and content workflows.

Fetch TikTok data through one operation-selectable Actor. Choose profiles, video details, comments, replies, user search, video search, hashtag videos, music metadata, music videos, captions or transcripts. This Actor returns structured JSON without requiring your own TikTok credentials. Content is resolved in the US region; alternative regions are not supported.

Quick start

Choose an operation, provide its required resource, and start with one page:

{
  "operation": "search-videos",
  "query": "cooking",
  "maxPages": 1
}

Use search-users to discover creators, comments to inspect a video's audience responses, or hashtag-videos with hashtag: "cooking" to explore a topic. For spoken text, choose captions first when existing subtitles are sufficient; choose transcript when you want automatic generation if subtitles are missing. Both require a public TikTok video url. An automatic request uses the automatic price even if existing captions satisfy it.

Inspect the Dataset JSON and run summary before increasing the workload. List operations return nested pages, not a flat spreadsheet row per video/comment.

Choose an operation

Operation	Required input	Output and billing unit
profile	username	One profile
profile-videos	username	One complete video page per result
video	url	One video detail result
comments	url	One complete comment page per result
replies	url, commentId	One complete reply page per result
music	musicId	One sound metadata result
search-users	query	One complete user page per result
search-videos	query	One complete video page per result
hashtag-videos	hashtag	One complete video page per result
music-videos	musicId	One complete video page per result
captions	url	One caption result, without generation
transcript	url	One automatic transcript result
videos	urls	One row per requested video; at most 20 URLs
transcripts	urls, optional mode	One row per URL; at most 10 URLs

Use a hashtag without its leading hash or spaces. Music and parent comment IDs must be numeric strings, not numbers. Usernames may include a leading at sign. Video URLs must use HTTPS and a TikTok hostname, including short share links.

Pagination and output

maxPages defaults to one and is capped at twenty. Every successful list request produces one dataset row whose data contains the complete array, has_more and next_cursor. Page prices do not multiply by the number of comments or videos in that array. Empty successful pages are also billed. This nested shape is easiest to export as JSON; do not interpret one row as one video on list operations.

Keep the operation, resource and query options unchanged when continuing with cursor. Source results can overlap and this Actor preserves those duplicates within page responses. A continuing cursor does not guarantee a populated next page. A failed page, including continuation 404, fails the run while preserving earlier rows; it is not presented as successful exhaustion. No complete-catalogue guarantee is made. OUTPUT reports completion, page-limit or budget termination for successful runs.

Captions and transcripts

captions and transcript batches with mode: "captions" never generate speech recognition. transcript and batches with mode: "auto" can generate a transcript when captions are unavailable. Generation supports videos up to ten minutes and may take up to three minutes. Captions and automatic transcripts have separate prices even when an automatic request finds existing captions.

Billing and failures

The Store pricing tab is authoritative. There is a start event, metadata-page and metadata-item events, and separate caption and automatic-transcript item events. Batch rows include success; failed items include status and error and are not charged as successful items. An entirely failed batch fails the run after saving its error rows. Duplicate URLs in batch input remain separate requested and billable entries.

The run budget is checked before dispatch. A batch may be shortened to fit it; inspect the completion summary and returned indexes. No automatic HTTP retry is performed. Do not resurrect a failed or migrated run: this Actor refuses to restart paid work within the same run. Start a new run only for results you still need; a new run has its own start and result charges. There is no atomic guarantee across external billing and dataset storage if the platform fails during delivery.

Media links in returned data are temporary upstream URLs. Download promptly if needed. This Actor does not provide durable storage, permanent downloads or a licence to reuse the media. Profile history with per-video rows remains available through the separate Influship TikTok Profile Scraper.