Skip to content

新增SOP: 抖音视频语音提取与内容分析流程#2

Draft
Homeless-Xu wants to merge 1 commit into
masterfrom
claude/voice-extraction-video-lysi4p
Draft

新增SOP: 抖音视频语音提取与内容分析流程#2
Homeless-Xu wants to merge 1 commit into
masterfrom
claude/voice-extraction-video-lysi4p

Conversation

@Homeless-Xu

Copy link
Copy Markdown
Owner

内容

◼︎3-Script/ 下新增《SOP-抖音视频语音提取分析.md》,沉淀本次验证过的完整流程:

  1. 短链解析:curl 跟随重定向拿到视频 ID
  2. 免登录取地址:抖音分享页(iesdouyin.com)内嵌 _ROUTER_DATA JSON,直接解析出真实播放地址与元数据(playwmplay 可去水印)
  3. 下载与预处理:curl 下载 mp4,ffmpeg 提取 16kHz 单声道 wav + 每 18 秒抽一帧(480px 宽,控制在 3~5 帧)
  4. 本地转写:faster-whisper(small 模型 / int8 / CPU),initial_prompt 喂主题词提高专有名词识别率
  5. AI 分析省 token 原则:以转写文本为主,少量帧画面为辅,硬字幕用于校对听错的专有名词

附常见坑对照表(yt-dlp 要 cookies、分享页改版定位、UA/Referer、其他平台走 yt-dlp 等)。

验证

流程已于 2026-07 在真实抖音视频(71 秒,ESP32 光剑项目)上端到端跑通:下载、抽帧、转写、分析全部成功。

🤖 Generated with Claude Code

https://claude.ai/code/session_01P35R3gxKDmafv2kku413jW


Generated by Claude Code

分享短链解析→分享页JSON提取真实地址→curl下载→ffmpeg提音频抽帧
→faster-whisper本地转写→AI分析(转写为主/少量帧为辅,省token)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01P35R3gxKDmafv2kku413jW
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants