wjs-voicedrop

Translates SRT subtitle files between languages with punctuation-bounded re-segmentation so cues end at real sentence breaks, supporting Simplified Chinese and English as first-class targets with bilingual SRT output.

Burns SRT subtitles into video using ffmpeg/libass for platforms that ignore embedded tracks (WeChat, Douyin), supports soft-mux togglable tracks, and handles full localization compositing (subs + dub + original audio bed) in a single encode pass.

A skill that transcribes audio/video into timestamped SRT subtitles, routing between Volcano ASR for Chinese and OpenAI Whisper for other languages, with word-level timestamps and punctuation-bounded cues.

A skill that generates time-aligned TTS voice dubs for videos from translated SRT subtitles, routing between Volcano TTS for Chinese and Edge TTS for other languages with optional multi-speaker support.

A post-production skill for video clips that adds AI-generated cover frames, HTML/CSS-quality captions synced to SRT, kinetic illustration overlays, chapter chips, and end-card CTAs — all composed in a single HyperFrames project and rendered in one final encode. Typically used downstream of the wjs-segmenting-video skill to produce upload-ready MP4s.

Turns a single English word into a self-contained HLS video lesson by stitching together every clip where the word is spoken from the mira video base, prepending a bilingual word-intro card with IPA transcription, Chinese gloss, and TTS audio, and appending a CTA card.

Converts video between horizontal and vertical orientations (e.g., 16:9 to 9:16) by cropping a narrow band that tracks the active speaker using MediaPipe face landmarks and mouth-aspect-ratio variance, keeping the talking person in frame for platforms like TikTok, YouTube Shorts, and WeChat Channels.

Orchestrates the end-to-end video localization pipeline by routing through four sub-skills: audio transcription, subtitle translation, dubbing, and subtitle burning into the final video.

Assists with writing and publishing WeChat Official Account (微信公众号) articles by lightly polishing drafts while preserving the author's voice, auto-generating cover images and explanation illustrations, and outputting content ready to paste into the WeChat backend.

Segments long-form videos with SRT transcripts into 3 to 6 stand-alone topical short clips by analyzing the transcript, selecting topic boundaries, and cropping for the target platform.