skills.rest
bibidu
bibiduCOMMUNITY
Explore all skills by this author →
GitHub

fengshui

Generates 9:16 vertical voiceover videos from Douyin links with local voice cloning.

Updated Jul 8, 2026
Search Tags:#text-to-speech#subtitles#ffmpeg#voice cloning#video generation#douyin#vertical video
One-click install
npx skills add https://github.com/bibidu/era --skill fengshui-bibidu
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fengshui
Source: https://github.com/bibidu/era/tree/main/.agents/skills/fengshui
Command: npx skills add https://github.com/bibidu/era --skill fengshui-bibidu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? It turns a single Douyin or video link into a finished 1080x1920 vertical voiceover video, handling transcript extraction, local voice cloning, title artwork, subtitles, encoding, and HTTPS delivery without manual editing. ## Core Features & Use Cases - Link-to-video pipeline: Extracts the Chinese voiceover from a Douyin link, rewrites under 5% of the wording, and produces a final narration script. - Local voice cloning with quality gates: Selects the highest-priority runnable engine (CosyVoice > VoxCPM2 > VoxCPM 0.5), clones an elderly voice from a fixed reference sample, and enforces word-level ASR, pause, loudness, pitch, and noise gates. - Cinematic visuals and titles: Generates a static 9:16 shanshui image, a two-column AI calligraphy title with the Jinyuan seal shown for 2 seconds, and bold Song-style subtitles synced to the narration. - Use Case: A user pastes a fengshui Douyin link and receives a verified HTTPS deep link to a playable 1080x1920 video after confirming a 10-second voice sample and the final first frame. ## Quick Start Use the fengshui skill to turn this Douyin link into a vertical voiceover video and give me the HTTPS preview link.

Frequently Asked Questions about fengshui

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn a Douyin link into a vertical voiceover video?▼

Provide the Douyin link and the skill extracts the Chinese voiceover, rewrites under 5% of the wording, clones an elderly voice locally, adds a shanshui background with calligraphy title and subtitles, and encodes a 1080x1920 MP4 delivered as an HTTPS preview link.

Which voice cloning engine does it use for Chinese narration?▼

It probes local engines in fixed priority order: CosyVoice, then VoxCPM2, then VoxCPM 0.5, selecting the highest-priority one that can safely load and pass quality gates. MiniMax is explicitly forbidden.

Can the video be generated fully automatically without review?▼

No. A mandatory confirmation gate requires the user to approve a 10-second real cloned audio sample and the final first frame before full production proceeds. Requests to skip confirmation or run fully automatic are not honored.

What audio quality checks does voice cloning require?▼

Cloned audio must pass word-level Chinese ASR matching, pause limits, loudness of -20 to -14 LUFS, True Peak under -1dBTP, pitch range checks, and noise or click audits. Any failure blocks video composition.

Why is an OSS raw link not accepted as video delivery?▼

Delivery must be a dedicated HTTPS deep link in the form ?tab=fengshui&v=<mp4 path> that binds the specific uploaded MP4. Raw OSS links, root paths, or generic tab links are forbidden because they do not prove the video plays.

What are the limitations of this video generation workflow?▼

It only produces 9:16 vertical voiceover videos from links, not multi-page graphic layouts, which are deprecated. It also requires local heavy workloads to run serially and will wait rather than run when machine resource prechecks fail.

Related Skills

Explore More Content & Communication Skills→
davepoondavepoon
3.4k

video-to-text

Extract Douyin video metadata and transcribe speech from local video files.

Community
Intermediate
claude-office-skillsclaude-office-skills
368

YouTube Automation

Automate YouTube video uploads, metadata optimization, scheduling, and analytics via MCP.

Official
Advanced
stevengonsalvezstevengonsalvez

video-frames

Extract video frames by timestamp or index using ffmpeg.

Community
Intermediate
missyouangeledmissyouangeled

tts

Convert text into speech audio using Kokoro offline or Noiz cloud backends.

Community
Advanced