🛠️ Fixed impersonation — the Chrome TLS anti-bot bypass was silently disabled (yt-dlp needs an ImpersonateTarget object, not the raw "chrome" string). Now active, cutting TikTok/Instagram/Facebook block rates.
🔄 Always-fresh extractors — the build now ships the yt-dlp nightly channel, fixing the recurring "Unable to extract webpage video data" caused by frozen/stale extractors.
⚡ Faster default run — set a 4 GB memory floor (more CPU) and trimmed the default example to short clips so the automated daily test reliably finishes within Apify's 5-minute window (the main cause of the actor flipping to "under maintenance").
🔧 Loosened curl_cffi pin so newer Chrome impersonation profiles are available.
0.1 — Initial release
🎬 Transcribe public TikTok, Instagram Reels, Facebook Reels, and YouTube Shorts with timestamps.
🔎 Automatic language detection (or force a specific language).
🛡️ Smart proxy ladder: Direct → Datacenter → Residential, sticky on Residential, with 3 residential retries.
💾 Live saving — every transcript is pushed to the dataset as soon as it is ready.
📊 Three output views: Overview, Transcripts, and Timestamped Lines.
