Tutorial Video Voiceover and Subtitle Composer

Problem

When producing tutorial videos, scripts, timing, TTS, subtitles, and final compositing are often split across manual steps. Hand-writing timecodes is tedious, while TTS credentials, ffmpeg, package versions, and template formats can interrupt the workflow. This skill organizes voiceover, subtitle, and compositing tasks into a resumable pipeline: the user supplies the video, script, and voice preference, while the assistant guides configuration and execution.

How it works

• Template authoring: Use subtitle-editor.html to drag in the video, press F to mark timing, adjust with ← and →, and generate a template prompt; a direct script_template.md file is also supported.
• Environment checks: The skill creates _scripts/, copies Python helpers, and checks for ffmpeg, websocket-client, and edge-tts, offering install guidance or completing package installation when needed.
• TTS selection: Supports iFlytek spark and edge-tts voices. iFlytek uses tts_config.json for credentials, while edge-tts maps choices such as Xiaoxiao, Yunxi, or Xiaoyi to neutral voices.
• Job execution: It generates job.yaml, then runs pipeline.py through cutting, TTS, alignment, subtitle generation, and compositing; failures can be retried using the emitted resume command.
• Subtitle handling: Supports frame-based MM:SS,FF and decimal MM:SS.mmm timing, converts math symbols into spoken forms, and can use polyphone or prosody markers in the template.

Boundaries

The skill requires a local ffmpeg installation and a compatible Python environment. iFlytek mode uses third-party credentials and should keep them out of version control. edge-tts depends on a package and online service, so network restrictions may require a mirror. Timecode commas in MM:SS,FF are frame separators, not decimal points. Before sharing, remove real app_id, api_key, api_secret, and private endpoints.