Back to Details

video-transcription

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Video Transcription

视频转录

Follow shared release-shell rules in:
  • postplus-shared
    release-shell rules
Use this skill when the input is video and the main problem is:
  • spoken transcript extraction
  • subtitle timing
  • edit-prep timestamps
  • multilingual caption generation
This skill is not a substitute for semantic video analysis.
请遵循以下共享的release-shell规则：
  • postplus-shared
    的release-shell规则
当输入为视频，且需要解决以下主要问题时，可使用此技能：
  • 提取语音转录文本
  • 字幕时间戳设置
  • 为视频剪辑准备时间戳
  • 生成多语言字幕
此技能无法替代语义视频分析。

Hosted Endpoint

托管端点

First-version hosted transcription endpoint:
  • hosted transcription capability
  • transcription-whisper-with-video
Use 
enableTimestamps=true
by default when the result will drive subtitles or edit decisions.
第一版托管转录端点：
  • 托管转录功能
  • transcription-whisper-with-video
当结果用于生成字幕或指导剪辑决策时，默认启用
enableTimestamps=true
。

Poll Behavior

轮询机制

Hosted video transcription is asynchronous. The script polls the prediction result URL until status is 
completed
or 
failed
. Default poll window: 150 attempts × 2 s = 5 minutes.
Real speech video typically completes within 30–60 s. Silent or very short videos may complete in under 10 s. If a job exceeds 5 minutes, the hosted provider is likely overloaded — retry rather than increasing the timeout further.
托管视频转录为异步操作。脚本会轮询预测结果URL，直到状态变为
completed
（完成）或
failed
（失败）。默认轮询窗口：150次尝试 × 2秒 = 5分钟。
包含真实语音的视频通常会在30–60秒内完成转录。无声音或极短的视频可能在10秒内完成。如果任务耗时超过5分钟，很可能是托管服务提供商负载过高——此时应重试，而非进一步延长超时时间。

Output Contract

输出约定

Persist:
  • request.json
  • response.json
  • manifest.json
  • provider output artifacts under 
    outputs/
需持久化保存以下文件：
  • request.json
  • response.json
  • manifest.json
  • 提供商输出的文件，存放于
    outputs/
    目录下

Scripts

脚本

  • scripts/transcribe_video.mjs
Polling support is shared with:
  • skills/40-creative/audio-transcription/scripts/poll_transcription.mjs
  • scripts/transcribe_video.mjs
轮询支持与以下脚本共享：
  • skills/40-creative/audio-transcription/scripts/poll_transcription.mjs

Read These References

参考文档

  • references/tool-contracts.md
  • references/normalized-transcript-schema.md
  • references/tool-contracts.md
  • references/normalized-transcript-schema.md

Release-Shell Execution Contract

Release-Shell执行约定

  • keep transcription requests, provider responses, normalized transcripts, and downloaded subtitle artifacts under 
    <work-folder>/.postplus/video-transcription/
  • keep only final user-facing transcript exports outside 
    .postplus/
  • start with a bounded first pass, usually one source file before larger batches
  • if hosted transcription capability is unavailable, unauthorized, or returns a stable network error, stop immediately instead of switching to ad hoc shell glue
  • 将转录请求、提供商响应、标准化转录文本以及下载的字幕文件存放于
    <work-folder>/.postplus/video-transcription/
    目录下
  • 仅将最终面向用户的转录文本导出文件存放在
    .postplus/
    目录之外
  • 先进行小规模测试，通常先处理一个源文件，再进行批量处理
  • 如果托管转录功能不可用、未授权或返回持续网络错误，请立即停止操作，不要切换到临时的shell脚本替代方案