What changed, and when.

Interface localisation for English, Spanish, Japanese and Khmer.

Publishing diagnostics tab reporting per-account OAuth health with a classified failure reason for every attempt.

Chrome profile isolation so separate publishing identities no longer share browser state.

TikTok authorisation now uses hex-encoded PKCE challenges as TikTok's desktop flow requires, fixing token exchange rejections.

Token exchange surfaces the platform's own error text instead of failing with an opaque key error.

Uploads interrupted by a crash are reset to pending on next launch rather than retried automatically.

Voice metadata commits that succeeded on the GPU but failed to persist are now replayed at startup before the window opens.

OAuth token encryption key permissions tightened to owner-only where the operating system supports it.

Voice Library with reusable cloned voice profiles built from a reference sample.

VoxCPM2 studio pipeline: sentence-boundary chunking, per-chunk quality checks, bounded retries and crossfade merging.

Loudness normalisation to −16 LUFS with a −1 dBFS peak ceiling across every provider.

Parallel voice batching to keep the GPU saturated across a batch of jobs.

Voice cache keys now include preset and speed, so preset changes correctly invalidate cached audio.

Voice drift and quality degradation on scripts longer than roughly two minutes.

Publishing Center as an independent module with its own isolated SQLite database.

YouTube, TikTok and Facebook publishers using each platform's official installed-application OAuth flow.

Throttled upload queue with retry, scheduling and a full per-attempt history.

Auto SEO metadata generation with per-platform profiles, scoring and constraint validation.

Folder watchers that pick up new exports automatically without starting a job on their own.

OAuth tokens encrypted at rest with Fernet using a locally generated key.

Director engine: beat, rhythm, emotion, motion, layout and colour-style planning.

Editor engine: attention tracking, continuity checking and semantic asset ranking.

Hardware encoder detection by real test encode across NVENC, QuickSync and AMF.

Pipeline checkpointing so an interrupted render resumes at its last completed stage.

Smart visual fallback chain guaranteeing every scene receives a visual.

Render profiles expanded to fast, balanced, high quality and lossless.

End-to-end pipeline: script, voice, footage, subtitles, render and export.

AI storyboard and scene planner driving shot selection and pacing.

ElevenLabs and OpenAI cloud voices with local Whisper subtitle transcription.

Batch generation across a list of topics with a shared footage cache.