Instagram Transcript Scraper

Turn any Instagram Reel or video post into clean, AI-generated text. Paste one or more Instagram URLs and get back the full spoken-audio transcript of each video — no login, no cookies, no browser extension. Built for content marketers, researchers, and AI agents that need Instagram video turned into searchable, repurposable text.

What it does

This actor takes Instagram Reel / video-post URLs (or bare shortcodes) and returns an AI-transcribed text of the spoken audio in each video. Carousel posts that contain multiple videos return one transcript per video. Photos, audio-less clips, and videos longer than the provider's 2-minute limit return an empty transcript with a clear explanation instead of failing.

It uses public data only. You never supply Instagram credentials, session cookies, or a sessionid — the actor routes requests through managed transcript providers (ScrapeCreators primary, SociaVault fallback) so it stays reliable even when one provider is down or rate-limited.

When to use it

• Repurpose Reels into blog posts, newsletters, captions, or scripts.
• Build a searchable archive of a creator's spoken content.
• Feed Instagram video content into an LLM / RAG pipeline as text.
• Accessibility: generate readable transcripts of video posts.
• Competitive and trend research across many Reels at once.

Not for: scraping whole profiles, hashtags, followers, comments, or photo captions. This actor transcribes the spoken audio of specific video posts. For profile, hashtag, or post-metadata scraping, use the matching Instagram actors in the same portfolio.

Output

One dataset record per input URL:

Field	Description
shortcode	Instagram post/reel shortcode (e.g. DZpQwxqimz2)
url	Canonical Instagram URL transcribed
transcript	Full spoken-audio transcript (carousel parts joined with blank lines); null when no speech
transcriptItems	Array of per-video parts { shortcode, mediaId, text } (length > 1 for carousels)
isCarousel	true when the post had multiple videos
language	Language hint label you passed in (metadata only)
source	Provider that produced the transcript (scrapecreators or sociavault)
status	ok, no_transcribable_audio, or provider_error
note	Human-readable explanation for non-ok records
scrapedAt	ISO 8601 extraction timestamp

Example record

{
  "shortcode": "DZpQwxqimz2",
  "url": "https://www.instagram.com/reel/DZpQwxqimz2/",
  "transcript": "We all know the story of Pompeii, right? ... This is the story I want to tell.",
  "transcriptItems": [{ "shortcode": "DZpQwxqimz2", "mediaId": "3920738671208852726", "text": "We all know the story of Pompeii..." }],
  "isCarousel": false,
  "language": "auto",
  "source": "sociavault",
  "status": "ok",
  "scrapedAt": "2026-06-26T03:10:00.000Z"
}

Input

{
  "instagramUrls": [
    "https://www.instagram.com/reel/DZpQwxqimz2/",
    "https://www.instagram.com/p/C9wYv3kSxJ-/"
  ],
  "maxTranscripts": 100,
  "language": "auto"
}

• instagramUrls (required) — Reel / video-post URLs or bare shortcodes.
• startUrls (optional) — alternative request-object form [{ "url": "..." }] for chaining from another actor.
• maxTranscripts (optional, default 100) — hard cap on transcribed-and-billed videos per run.
• language (optional, default auto) — metadata label; transcription auto-detects language regardless.

Pricing

Pay-per-event:

• Actor start — $0.00005 per run.
• Transcript extracted — $0.01 per video successfully transcribed (one charge per post; carousel items share the charge).

You are only charged for videos that produce a transcript. Photos, no-speech clips, over-length videos, and provider errors are not billed for the transcript event. Each run prints an upfront cost cap and a final billed amount in the log and status message.

How it works

1. Each input is normalized to an Instagram shortcode + canonical URL (duplicates removed).
2. For each target, the actor calls ScrapeCreators' Instagram transcript endpoint first (cheapest). If it is out of credits, rate-limited, errors, or returns no transcript, it falls back to SociaVault.
3. Both providers run AI transcription on the video audio (typically 10–30 seconds per video). Videos must be under ~2 minutes; if no speech is detected the transcript is null.
4. Results are pushed to the dataset; a run summary is written to the OUTPUT key-value record.

Reliability notes

• No cookies required — public data only via managed providers.
• Graceful degradation — non-video posts and no-speech clips return a labeled empty record, not a crash.
• Two-provider redundancy — one provider outage does not break the run.
• Honest failure — if every request fails at both providers (e.g. total provider outage), the run is marked failed with a clear message rather than silently succeeding empty.

MCP / AI-agent ready

Single-URL-in, structured-JSON-out. Discoverable through Apify MCP as apify--instagram-transcript-scraper. Input descriptions and output fields are written for LLM tool-selection.

Legal

Use this actor only for content you are permitted to access and process. It extracts publicly available Instagram video data. You are responsible for complying with Instagram's Terms of Service, applicable copyright, and data-protection laws (including GDPR/CCPA) in your jurisdiction. This actor does not access private accounts or login-gated content.