# Changelog of Best Tiktok Ai Transcript Extractor (`sian.agency/best-tiktok-ai-transcript-extractor`) Actor

- **URL**: https://apify.com/sian.agency/best-tiktok-ai-transcript-extractor/changelog.md
- **Full Actor documentation**: https://apify.com/sian.agency/best-tiktok-ai-transcript-extractor.md

## Changelog

All notable changes to **TikTok AI Transcript Extractor** will be documented in this file.

### \[2026-08-24]

#### 🛡️ Version 2.2 — You Only Pay For Videos That Return Data

- **🔇 Silent videos are free** - a TikTok with no spoken audio (photo slideshows, music-only clips, ambient footage) is no longer billed as a transcription. You are charged for videos that come back with data, never for the ones that cannot be read
- **🎯 No more invented transcripts** - speech recognition engines famously hallucinate "Thanks for watching!" over silence. Those are now detected and blanked, so an empty transcript field means genuinely no speech rather than a plausible-looking fabrication you would have to catch yourself
- **📦 Full rows on quiet videos** - a video with no speech still returns its caption, cover image, creator profile, engagement metrics, hashtags, music and location. You keep all 45 data fields whether or not there was anything to transcribe
- **🔁 Tougher fallback for stubborn videos** - when TikTok answers with a stub that carries only a video ID, the actor now rejects it and re-sources the video instead of shipping you an almost-empty row

#### 📄 New Run Report

- **📊 Every run explains itself** - the run report now leads with what you got (videos transcribed, words captured, minutes of audio), gives you a one-click copy of the full transcript text, and itemizes exactly what each run cost
- **🔧 Failed URLs come with the fix** - each URL that did not work is listed with the specific reason and what to change, plus a one-click retry of just those URLs
- **⛔ Short runs say why** - if a run stops before finishing your list, the report names the reason and how many videos were left untouched, instead of quietly showing a smaller number

#### 💎 User Benefits

- **Predictable costs:** the transcription charge tracks videos that actually returned data, so bulk runs over mixed content stop paying for silence
- **Trustworthy data:** an empty transcript now means empty, which matters when you are feeding thousands of rows into a search index or an AI pipeline
- **Faster triage:** the run report turns a failed batch into a list of URLs with fixes attached

#### 🎯 Use Cases

- Recipe and product researchers running bulk TikTok batches where a third of the videos are music-only, and needing the caption, pinned comment and creator data from those too
- Social media analysts building transcript databases for AI trend detection, where a fabricated "Thanks for watching" line poisons the index
- Agencies reconciling monthly Apify spend against delivered rows for client reporting

### \[2026-06-24]

#### 💬 Version 2.1 — Pinned & Creator Comments (opt-in add-on)

- **📌 Grab the pinned comment** - new `includeComments` toggle extracts the comments a creator pinned to the top of a video, plus the creator's own (unpinned) comments. When a video is music-only and the caption is empty, the full recipe, links, or key details often live in a pinned comment — now you capture them too
- **🧱 Structured output** - a full `comments` array (pinned comments first, then top comments) plus ready-made `pinnedComments` and `creatorComments` shortlists; each entry carries the comment text, author, like count, and pinned/creator flags

#### 💰 Pricing

- The comments add-on is available on **paid Apify plans**, billed per video **only** when the toggle is enabled **and** at least one comment is returned. Transcript-only runs are completely unaffected — no extra cost

#### 💎 User Benefits

- **Capture what's not spoken:** recipes, product links, and details creators put only in a pinned comment
- **Pay as you go:** opt-in per run, charged only when comments are actually found

### \[2026-06-05]

#### 🚀 Version 2.0 — Faster, Smarter, Multilingual!

- **🌍 English Translation Built In** - New `translateToEnglish` toggle converts any spoken language to an English transcript — segments and SRT/VTT subtitles included. Analyze Spanish, German, Japanese, or Korean TikTok content in one searchable English database
- **📥 Watermark-Free Download URLs** - Every result now includes a clean, directly usable video download link — perfect for content archiving and editing workflows
- **⚡ 3x Faster Processing** - Re-engineered extraction pipeline delivers transcripts in seconds, not minutes
- **🛡️ Built-In Redundancy** - Automatic fallback data sources keep your success rate high even when TikTok changes things under the hood
- **🔗 Smarter Short-Link Handling** - vm.tiktok.com and vt.tiktok.com share links resolve more reliably across all regions

#### 💎 User Benefits

- **Go global:** Research international creators and viral trends without language barriers
- **Save hours:** Faster runs mean bulk jobs of 100+ videos finish dramatically sooner
- **Fewer failures:** Redundant data sourcing means fewer retries and more complete datasets
- **Cleaner assets:** Watermark-free video URLs ready for repurposing and editing

#### 🎯 Use Cases

- Global brand teams monitoring TikTok trends across US, LATAM, and European markets in one English-language dashboard
- Content agencies extracting viral scripts from international creators for cross-market campaign inspiration
- Social media analysts building multilingual transcript databases for AI-powered trend detection

### \[2026-04-05]

#### ✨ New Features

- **SRT & VTT Subtitles:** Auto-generated SRT and WebVTT subtitle formats from transcript segments — ready for video editors and media players
- **Speaker Diarization & Word-Level Timestamps:** Segments now include `speaker`, `language`, and a `words` array with per-word timing, confidence scores, and speaker labels
- **Creator Profile Data:** Added `followerCount`, `authorHeartCount`, `authorVideoCount`, `authorBio`, and `authorVerified` fields
- **Engagement Expansion:** Added `collectCount` (bookmarks/saves) to engagement metrics
- **Content Classification:** Added `diversificationLabels` (content categories), `textLanguage` (detected language), and `isAd` (advertisement detection)
- **Location Data:** Added `locationCreated` (country code), `poiName`, and `poiAddress` for geo-tagged videos
- **Video Metadata:** Added `fileSize` and `scrapingDate` fields
- **Schema Additions:** `coverImageUrl` now properly defined in dataset schema fields

#### 🐛 Bug Fixes

- **Video Duration Fix:** `videoDuration` now correctly shows seconds (e.g. 50) instead of being divided by 1000 (was showing 0.05)

#### 🎯 User Benefits

- **Richer Analysis:** 45+ data fields (up from 30+) for deeper content insights
- **Speaker Identification:** Know who said what in multi-speaker TikTok videos
- **Creator Intelligence:** Full creator profile data for influencer analysis and competitor monitoring
- **Geo Insights:** Location-based analysis for regional content trends
- **Content Categorization:** TikTok's own content labels for automated topic classification

### \[2025-11-08]

#### ✨ New Features

- **Timestamped Segments Field:** Added `segments` array containing transcript broken into timestamped segments with id, text, start, and end times for precise content analysis
- **Improved Input Schema:** Enhanced formatting with better line breaks between Features and Use Cases sections

#### 🎯 User Benefits

- **Better UX:** Most important data visible first in results (transcript → segments → cover image → metrics)
- **Easier Onboarding:** Clear, concise documentation without fluff
- **Professional Output:** Clean dataset structure optimized for analysis
- **Precise Timestamps:** Segments field enables exact time-based content navigation and analysis