Instagram Transcript Scraper & Reels Monitor

Convert public Reels into searchable transcripts, subtitles, hooks, keywords, offers, promo codes, and monitoring events.

Transcribe public Instagram Reels or recent profile videos, monitor only new Reels, detect spoken keywords, hooks, offers, promo codes, and calls to action, and export JSON, SRT, and VTT.

All examples below use fictional DEMO-* identifiers, reserved .test domains, and masked contact details. They do not represent real people, listings, products, or companies.

🚀 What this Actor does

• Flexible collection: urls, profiles, monitor.
• Structured output: clean JSON records organized into transcripts, monitoring, subtitles views.
• Production-ready: concurrency, retries, proxy support, limits, and source-aware diagnostics are exposed through the Actor input.
• Easy automation: run from the Apify Console, API, schedules, webhooks, Make, Zapier, or any HTTP client.

🎛️ Modes and workflows

Mode or workflow	What it does
urls	Transcribe exact public Reel or video URLs.
profiles	Discover recent public Reels from each profile and transcribe the selected items.
monitor	Reuse a monitor ID and process only newly discovered Reels after the initial snapshot.

💡 Common use cases

• Creative and hook research
• Spoken keyword, offer, and promo-code monitoring
• Searchable subtitle and transcript archives

📥 Input schema

Configure the Actor in the Input tab or send the same JSON through the API. Fields not needed for your workflow can be omitted.

Field	Type	Description	Default
mode required	string	Choose direct Reel URLs, multi-profile research, or stateful monitoring. Allowed: urls, profiles, monitor.	profiles
urls	array	Used only in Reel URLs mode. Supports public /reel/, /p/, and /tv/ links or bare shortcodes.	—
profiles	array	Used in Profile research and Monitor modes. Accepts usernames, @usernames, or full profile URLs.	["nasa"]
maxReelsPerProfile	integer	Inspect between 1 and 12 recent public Reels per profile. Keep 1 for the fastest large-list runs.	1
firstTranscriptPerProfile	boolean	When enabled in Profile research mode, inspect recent Reels newest-first and stop after the first Reel with real speech. This produces at most one paid result per submitted profile instead of losing profiles whose newest Reel is silent.	false
monitorId	string	Stable name such as competitor-beauty-us. Reuse it on every scheduled run. When empty, the profile list becomes the ID.	—
firstRunMode	string	Process current Reels immediately, or save them only as a baseline so later runs return newer posts. Allowed: processCurrent, baselineOnly.	processCurrent
spokenKeywords	array	Optional case-insensitive words or brand names. Matches include timestamps and surrounding speech.	—
hookSeconds	number	Return the spoken opening from this many seconds of the Reel.	5
language	string	Automatic detection is recommended. Force a language only when you know it in advance. Allowed: auto, en, es, fr, de, it, pt, ru, uk, pl, tr, ar, hi, ja, ko, zh.	auto
preferNativeCaptions	boolean	Use a genuine timed subtitle track when Instagram exposes one; otherwise automatically fall back to speech-to-text.	true
speechRecognitionEngine	string	Speech-recognition selection. For production transcription, configure Deepgram Nova-3 with deepgramApiKey.	auto
deepgramApiKey	string	Encrypted by Apify. Recommended for bulk runs: Nova-3 accepts Instagram CDN URLs directly and supports up to 50 concurrent pre-recorded requests on eligible plans.	—
maxDurationSeconds	integer	Skip longer videos before speech recognition. This keeps run cost predictable; skipped videos are saved separately and are not billed as transcript results.	180
proxyConfiguration	object	Default connection settings are preconfigured for reliable public Instagram requests.	{"useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"], "apifyProxyCountry": "US"}
maxConcurrency	integer	Media downloads and recognition jobs processed in parallel. Local mode automatically caps this to available CPUs.	8
transcriptionConcurrency	integer	Cloud recognition can process several Reels at once; local mode automatically caps workers to available CPUs.	8
maxRetries	integer	Retry temporarily blocked, rate-limited, or interrupted Instagram requests.	3
requestTimeoutSecs	integer	Maximum time allowed for one Instagram or media request.	35

▶️ Example input

{
  "mode": "profiles",
  "profiles": [
    "demo_creator"
  ],
  "maxReelsPerProfile": 10,
  "language": "auto",
  "spokenKeywords": [
    "launch",
    "discount"
  ],
  "deepgramApiKey": "<YOUR_DEEPGRAM_API_KEY>"
}

🔐 Store deepgramApiKey as an encrypted input. Deepgram Nova-3 is the documented speech-recognition path for this Actor; only audio that requires transcription consumes Deepgram usage.

📦 Output schema

Each successful item is written to the default dataset. Select a dataset view in the Apify Console or export the full dataset as JSON, CSV, Excel, XML, or RSS.

transcripts — Reels & transcripts

Transcript and speech-analysis records.

Fields: coverUrl, url, authorUsername, postedAt, viewCount, durationSeconds, transcriptStatus, transcriptionSource, transcriptLanguage, hookText, matchedKeywords, keywordMatchCount, ctaText, offerMentions, promoCodes, transcript

monitoring — Monitoring

New-item and monitoring metadata.

Fields: recordType, mode, sourceProfile, isNew, shortcode, url, authorUsername, postedAt, matchedKeywords, keywordMentions, offerMentions, promoCodes, scrapedAt

subtitles — Subtitle exports

SRT, VTT, and segment-level subtitle output.

Fields: url, authorUsername, transcriptLanguage, transcriptSrt, transcriptVtt, transcriptSegmentsJson

📤 Example output

{
  "url": "https://www.instagram.com/reel/DEMO-REEL-01/",
  "authorUsername": "demo_creator",
  "durationSeconds": 27.4,
  "transcriptStatus": "completed",
  "transcriptionSource": "deepgram_nova_3",
  "transcriptLanguage": "en",
  "hookText": "Here is the fastest way to test the idea.",
  "matchedKeywords": [
    "launch"
  ],
  "promoCodes": [
    "DEMO20"
  ],
  "transcript": "Demo transcript text for documentation."
}

Missing source values are returned as null, empty arrays, or documented availability/status fields; the Actor does not invent unavailable source data.

🔌 API example

Replace the placeholder with an Apify token and send the same input used in the Console:

curl -X POST "https://api.apify.com/v2/acts/trakk~instagram-transcript-scraper-reels-monitor/runs?token=<YOUR_APIFY_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"mode": "profiles", "profiles": ["demo_creator"], "maxReelsPerProfile": 10, "language": "auto", "spokenKeywords": ["launch", "discount"], "deepgramApiKey": "<YOUR_DEEPGRAM_API_KEY>"}'

For synchronous integrations, use the run-sync-get-dataset-items API endpoint. For larger jobs, start an asynchronous run and consume its default dataset when the run succeeds.

🧩 Automation and exports

• Schedule recurring runs from the Apify Console.
• Trigger downstream systems with webhooks when a run succeeds or fails.
• Reuse named monitoring keys or stores where the selected workflow supports change tracking.
• Export dataset views to JSON, CSV, Excel, XML, or RSS, or access items through the Apify API.

❓ FAQ

Do I need a login or browser session?

Use only the inputs shown in the Actor schema. If authentication or cookies are supported, the relevant encrypted field is explicitly available in the Input tab; otherwise no account is required.

Why can some fields be empty?

Source pages vary by region, content type, privacy settings, and availability. Optional enrichment also depends on the selected mode. Empty values are preserved honestly instead of being guessed.

How should I run this at scale?

Start with a small representative input, inspect the dataset and cost, then raise item limits and concurrency gradually. Use Apify Proxy when the schema exposes it, and use schedules plus monitoring keys for recurring collection.

Is the example data real?

No. Every example in this README is intentionally fictional and uses demo identifiers, reserved domains, or masked contact values.