Instagram Reels Transcript Scraper — Reel to Text
Pricing
from $1.40 / 1,000 videos
Instagram Reels, TikTok and YouTube Shorts to text: transcript with timestamps, on-screen text, the 3-second hook, likes, views and a viral score vs the author's usual. No login, no cookies. Failed and private videos are free.
type
Optional
Row kind: 'reel' for a processed video, 'error' for a URL that could not be processed. Error rows are never billed.
url
Optional
The video URL exactly as submitted.
platform
Optional
Source platform detected from the URL: instagram, tiktok or youtube.
videoId
Optional
Platform video id.
durationSeconds
Optional
Video length in seconds.
views
Optional
Plays of this video. TikTok and YouTube only — Instagram hides views from logged-out visitors.
likes
Optional
Likes.
caption
Optional
Post caption text.
postedAt
Optional
Publish time, ISO 8601 UTC.
isOutlier
Optional
True when the Reel got at least 3× the author's usual views.
language
Optional
Language of the speech, detected from the audio. Null when the video has no usable speech.
contentClass
Optional
What kind of material this video is, decided from measurements: narrated (the voice carries the script), borrowed_audio (a trending sound or someone else's clip, not the author's script), silent (no speech, meaning lives in the on-screen text), empty (nothing analysable).
contentClassReason
Optional
Why that class was chosen, in plain words.
speechDetected
Optional
Whether the audio contains real speech. Speech-recognition models invent text over music, so this is decided from word density and how much of the clip is covered, not from the model's own confidence.
speechCoverage
Optional
Share of the video occupied by recognised speech, 0 to 1. Around 0.9 for a talking head, under 0.15 for music.
transcriptText
Optional
Full spoken transcript as one string.
transcript
Optional
Transcript split into timestamped segments. Present when timestamped segments are enabled.
firstThreeSecondsSpeech
Optional
Exactly the words spoken before the 3.0 second mark, cut on word boundaries — the spoken hook.
onScreenText
Optional
Text written on screen for the viewer: title cards, captions the author typed, calls to action. Burned-in subtitles that merely repeat the speech are excluded.
onScreenTextJoined
Optional
The same on-screen text as one newline-separated string, for spreadsheets.
firstThreeSecondsOnScreen
Optional
On-screen text shown before the 3.0 second mark — the visual hook.
burnedInCaptionCount
Optional
How many on-screen text tracks were identified as subtitles duplicating the speech and therefore excluded.
screenRecording
Optional
True when the video is a screen recording, where most recognised text belongs to an application interface rather than to the author.
errorCode
Optional
Short machine-readable failure reason on error rows: download_failed, timeout, unsupported_url, audio_failed, transcription_failed, ocr_failed, processing_failed.
errorMessage
Optional
Human-readable failure reason on error rows.
scrapedAt
Optional
UTC timestamp when the video was processed, ISO 8601.
vsAuthorMedian
Optional
This video divided by the author's median. 5 means five times the author's usual.
vsAuthorBasis
Optional
What was compared: views or likes.
comments
Optional
Comments.