TikTok Transcript & Instagram Reels Transcript (+ YouTube Shorts)

Get a TikTok transcript, Instagram Reels transcript, or YouTube Shorts transcript as clean text, timestamped segments, or an SRT subtitle file, from a plain list of video URLs. Mix all three platforms in one run. No login, no cookies, no API key.

Built for video-to-text pipelines: AI/RAG ingestion, content repurposing (turn a Reel into a blog post or newsletter), competitor and trend research, and subtitles for re-uploads. Works the same when called by an AI agent over Apify's MCP server.

Why it's cheap

Most transcript tools run AI speech-to-text on every video. But TikTok and YouTube already publish captions for most videos. This Actor reads those captions first (a few KB, no audio download) and only runs AI transcription when a video has none. You pay for what was actually needed, at your Apify discount tier (set by your Apify plan):

What happened	Free	Bronze & Silver	Gold and up
TikTok video with captions	$0.0025	$0.002	$0.0015
YouTube Short with captions	$0.004	$0.004	$0.004
No captions, so AI transcribed the audio (all Instagram Reels, some TikToks), per started minute of video	$0.008	$0.008	$0.008
Failed video, bad URL, or no speech found	free	free	free

Each run also has a start fee of $0.0003 per GB of memory (up to 1 GB counts once). With AI fallback on (the default) a run uses 4 GB, so $0.0012 per run; with aiFallback off it uses 1 GB, so $0.0003. Batch your URLs to spread it.

Example: 100 TikToks with captions in one run on the Free tier cost 100 x $0.0025 + $0.0012 = $0.25.

What you get (one item per URL)

{
    "url": "https://www.tiktok.com/@nasa/video/7689547576265149710",
    "platform": "tiktok",
    "videoId": "7689547576265149710",
    "author": "nasa",
    "durationSec": 121,
    "language": "eng-US",
    "source": "captions",
    "text": "Full transcript as one string...",
    "segments": [{"start": 0.0, "end": 2.4, "text": "First caption line"}],
    "srt": "1\n00:00:00,000 --> 00:00:02,400\nFirst caption line\n",
    "output": "Full transcript as one string...",
    "success": true
}

• source: captions (the platform's own captions), ai (AI speech-to-text), or none (nothing found; see error).
• text, segments and srt are always included. output repeats the one you picked in outputFormat, for tools that want a single field.
• A bad or deleted URL gets an item with success: false and an error; it never stops the rest of the run.

Input

{
    "urls": [
        "https://www.tiktok.com/@nasa/video/7689547576265149710",
        "https://www.youtube.com/shorts/DBhAROQM51E",
        "https://www.instagram.com/reel/DduQxrBjnsB/"
    ]
}

Field	Default	Notes
urls	required	TikTok, Instagram Reel/post, or YouTube Shorts URLs
language	"auto"	auto = each video's original language. Or an ISO code (en, es, pt...) to prefer that caption track; YouTube can translate into it
aiFallback	true	Transcribe the audio with AI when a video has no captions
localModelSize	"base"	AI model: base (more accurate) or tiny (a bit faster). Same price
outputFormat	"text"	Which of text / segments / srt goes into output
proxyConfiguration	Apify residential	Used only when a platform blocks direct requests

How it works

1. Captions first. TikTok's own captions (present on most talking videos) and YouTube's captions, original language by default.
2. AI fallback. No captions (Instagram never publishes them)? The Actor downloads the smallest audio track and transcribes it with Whisper (faster-whisper, running inside the Actor; by default your audio isn't sent to any third-party API). Music-only clips come back empty and free instead of with made-up text.
3. Cheapest route that works. TikTok and Instagram are fetched directly; the residential proxy is used for YouTube and as a retry, which keeps prices low.

Tips

• Batch your URLs: one run with 100 URLs is faster and cheaper than 100 runs.
• For English subtitles of non-English YouTube Shorts, set language to en.
• Set a maximum cost per run in Apify; the Actor stops cleanly when it's reached and never starts an AI transcription it can't finish within it.

Using it from an AI agent / MCP

Available through Apify's MCP server like any Store Actor. An agent passes a list of URLs and gets structured transcripts back.

Limits

• Private, deleted, age-restricted or region-locked videos can't be read.
• Instagram occasionally rate-limits; failed items say so and cost nothing.
• AI transcripts are good on clear speech and weaker over loud music or overlapping voices.

Running locally

python -m venv venv && source venv/bin/activate
pip install -r requirements.txt   # plus ffmpeg for the AI path

mkdir -p storage/key_value_stores/default
echo '{"urls": ["https://www.youtube.com/shorts/DBhAROQM51E"]}' > storage/key_value_stores/default/INPUT.json
APIFY_LOCAL_STORAGE_DIR=./storage python -m src

Results land in storage/datasets/default/.

---

Not affiliated with or endorsed by TikTok/ByteDance, Instagram/Meta or YouTube/Google. Trademarks belong to their owners.