🎙️ Instagram Reels Transcript — Speech to Text with Whisper

$0.02 per Reel with speech · silent Reels are free · 🔒 no login, no cookies, no proxy setup · 🌍 99 languages, auto-detected 🔗 View on Apify Store · 🇺🇸 English · 🇧🇷 Português

You paid for a "transcript scraper", the run came back with URLs and transcript: null, and Instagram never actually had a transcript to give. This actor does not wait for Instagram: it downloads the Reel's own audio and transcribes it with Whisper running inside the Actor — the same engine as our Video Transcriber, no external API, no key, no quota.

Give a profile (newest N Reels) or a list of Reel URLs and get, per Reel, the spoken words, timestamped segments, an SRT, the detected language — plus caption, duration, plays, likes and comments.

Who it's for: social-media and content teams mining what creators say, brand and competitor monitoring, influencer vetting, ad and UGC research, RAG and search over short video.

▶️ Click Try for free — the prefilled profile returns real transcripts in about a minute.

⚖️ What changed in 2.0

Before (1.x)	Now (2.0)
Read the transcript Instagram attaches to a Reel — logged-out Instagram exposes none; runs returned url_only	Whisper transcribes the Reel's audio — every Reel with speech gets a transcript
Browser + proxies, $0.002 per Reel plus compute and proxy billed to you	No browser; $0.02 per Reel with speech, platform usage included
Metadata-only rows were billed	Rows without speech, unavailable media and provider errors are free

✨ Features

• 🎙️ Real speech-to-text — faster-whisper (small by default; base/tiny for speed), vad_filter on, language auto-detected or forced.
• 📋 Profile or URL input — newest maxReels Reels of each public profile, or exact Reel URLs / shortcodes (up to 200).
• ⏱️ Timestamps + SRT — segments[] with start/end seconds and a ready srt, plus wordCount.
• 📊 Reel metadata — caption, hashtags, duration, plays, likes, comments, publish date, thumbnail, paid-partnership flag.
• 🛡️ Honest empties — music-only Reels (or a one-word fragment) come back as has_transcript: false, errorCode: NO_SPEECH, not charged.
• 💳 Spend control — the actor never requests more Reels than your maximum charge per run allows.

📥 Input

Field	Type	Default	Description
profiles	array	["cristiano"]	Usernames or profile URLs (public).
reelUrls	array	—	Reel URLs / shortcodes (max 200).
maxReels	integer	10	Newest Reels per profile (1–100).
language	string	auto	ISO 639-1 code to skip detection.
model	enum	small	tiny · base · small.
maxSecondsPerReel	integer	180	Longer Reels are cut here and flagged truncated.
includeSrt	boolean	true	Add srt.
brightDataApiToken	secret	—	Optional: your own provider token for Reel discovery.

📤 Output (per Reel)

{
  "url": "https://www.instagram.com/reel/DbDp7T4olyC/",
  "shortcode": "DbDp7T4olyC",
  "owner_username": "cristiano",
  "caption": "Greatness takes many forms 💪 …",
  "video_duration": 84.3,
  "play_count": 18400000,
  "like_count": 2544479,
  "comment_count": 25063,
  "published_at": "2026-07-21T13:55:36.000Z",
  "has_transcript": true,
  "transcript_language": "en",
  "transcript": "Greatness is not a moment, it is a habit…",
  "wordCount": 118,
  "transcribed_seconds": 84.3,
  "truncated": false,
  "segments": [{ "start": 0.0, "end": 3.4, "text": "Greatness is not a moment," }],
  "srt": "1\n00:00:00,000 --> 00:00:03,400\nGreatness is not a moment,\n",
  "source": "whisper",
  "_dataQuality": "transcript"
}

Reels without speech: has_transcript: false, errorCode: "NO_SPEECH", _dataQuality: "no_speech" (free). Media that could not be fetched: errorCode: "MEDIA_DOWNLOAD_FAILED" (free). Private profile / wrong username: one NO_REELS_FOUND diagnostic row (free).

🎯 Use cases

• Pull the talking points of a creator's last 50 Reels into a spreadsheet or an LLM.
• Monitor what influencers say about your brand, not just what they write in the caption.
• Vet influencers for tone, claims and language before a deal.
• Build subtitle files (srt) for repurposed clips.
• Feed Reel speech into search, RAG and trend analysis alongside our TikTok and YouTube actors.

✅ Capabilities & Limits

• Public profiles and public Reels only; private accounts return NO_REELS_FOUND.
• Whisper quality: small handles Portuguese, Spanish and English well; heavy music beds reduce accuracy; on-screen text is not read (only speech).
• Each Reel is transcribed up to maxSecondsPerReel (default 3 min); longer Reels are cut and flagged.
• Reel discovery comes from a third-party provider — expect 30–90 s before the first transcript appears; large profiles (100 Reels) take several minutes.
• Speaker labels and word-level timestamps are not produced.
• Throughput on the default 2 GB ≈ 0.5× real time: a 30-second Reel takes ~15 s of compute; give the run 4 GB to halve the wall time of big batches.

❓ FAQ

Q: Do I need cookies or an Instagram session? A: No. Discovery and audio come from public data; nothing is logged in on your behalf.

Q: Why did 1.x return transcript: null? A: Instagram only shows auto-captions inside its own logged-in apps; nothing is exposed publicly. 2.0 stopped depending on it.

Q: A Reel came back NO_SPEECH but people are talking in it. A: Whisper's voice filter can skip speech buried under loud music. Re-run with model: small (default) and, if you know it, set language. You are not charged for NO_SPEECH rows.

Q: Can I transcribe TikToks or YouTube videos too? A: Use Video Transcriber (same engine, any video URL) or YouTube Transcript Scraper (YouTube captions, $0.01/video).

💰 Pricing

Pay per event — $0.02 per Reel with a transcript (reel-transcripted). Reels without speech, unavailable media, provider errors and diagnostics are free. Platform usage (compute, discovery provider) is included in the price. Runs stop automatically at your maximum charge per run.

🔗 Related actors

• Video Transcriber — Whisper for TikTok, YouTube, Instagram and direct video links, per minute.
• YouTube Transcript Scraper — YouTube captions to text/SRT.
• Instagram Reels Scraper — Reel metrics without transcription.

📝 Changelog

• 2.0.0 (2026-08-29) — Rebuilt on Whisper: Reel audio transcribed inside the Actor; profile and URL input; segments + SRT; $0.02 per Reel with speech, platform usage included; silent Reels free. The 1.x native-transcript approach was retired because logged-out Instagram exposes no transcript.
• 1.1 (2026-08) — native-transcript attempt (deprecated).

---

🇧🇷 Português

Você pagou por um "transcript scraper", o run voltou com transcript: null — e o Instagram nunca teve transcrição pública para entregar. Este actor não depende do Instagram: baixa o áudio do próprio Reel e transcreve com Whisper rodando dentro do Actor (mesmo motor do nosso Video Transcriber). Sem API externa, sem chave, sem login.

Informe um perfil (os N Reels mais recentes) ou URLs de Reels e receba, por Reel, a fala transcrita, segmentos com timestamps, SRT e idioma — mais legenda, duração, plays, curtidas e comentários.

✨ Recursos

• Transcrição real por Whisper (small por padrão), idioma automático ou forçado.
• Entrada por perfil público ou por URL/shortcode de Reel (até 200).
• segments[], srt, wordCount e metadados do Reel.
• Reels só com música voltam como has_transcript: false / NO_SPEECH e não são cobrados.
• O actor nunca pede mais Reels do que o seu limite de gasto por run permite.

💰 Preço

$0,02 por Reel com transcrição (reel-transcripted). Reels sem fala, mídia indisponível, erros do provedor e diagnósticos não são cobrados. Uso da plataforma (computação e descoberta) incluído no preço.

❓ FAQ

Precisa de cookies? Não. Por que a 1.x devolvia transcript: null? O Instagram só mostra legendas automáticas dentro dos apps logados; nada é exposto publicamente. A 2.0 deixou de depender disso. TikTok ou YouTube? Use o Video Transcriber ou o YouTube Transcript Scraper.

📝 Changelog

• 2.0.0 (2026-08-29) — reconstruído com Whisper; $0,02 por Reel com fala; Reels silenciosos grátis.