openclaw-voc-skill/openclaw-skills/video-creation/transcript-to-video/api-config.json

{
  "name": "transcript-to-video",
  "displayName": "逐字稿 AI 视频生成工作流",
  "description": "输入逐字稿文本，AI 自动拆分分镜脚本，批量生成视频素材，输出视频片段 URL 列表和 SRT 字幕文件。支持文生视频和先图后视频两种模式。",
  "category": "video-creation",
  "version": "1.0.0",
  "type": "orchestration",
  "parameters": {
    "type": "object",
    "required": [
      "transcript"
    ],
    "properties": {
      "transcript": {
        "type": "string",
        "description": "逐字稿全文（中文），Agent 将自动拆分为分镜段"
      },
      "segmentDuration": {
        "type": "number",
        "description": "姣忔鐩爣鏃堕暱锛堢锛夛紝褰卞搷鍒嗛暅鎷嗗垎绮掑害",
        "default": 15
      },
      "videoModel": {
        "type": "string",
        "enum": [
          "720p",
          "1080p",
          "pro"
        ],
        "description": "瑙嗛鐢熸垚妯″瀷璐ㄩ噺绛夌骇",
        "default": "720p"
      },
      "videoMethod": {
        "type": "string",
        "enum": [
          "text2video",
          "img2video"
        ],
        "description": "text2video=纯文生视频（更快），img2video=先生成图片再图生视频（效果更可控）",
        "default": "text2video"
      },
      "aspectRatio": {
        "type": "string",
        "enum": [
          "16:9",
          "9:16",
          "1:1",
          "4:3",
          "3:4",
          "21:9"
        ],
        "description": "瑙嗛鐢婚潰姣斾緥",
        "default": "16:9"
      },
      "frames": {
        "type": "integer",
        "enum": [
          121,
          241
        ],
        "description": "视频帧数，121=5秒，241=10秒",
        "default": 121
      }
    }
  },
  "pipeline": [
    {
      "step": 1,
      "name": "AI鍒嗛暅鑴氭湰鐢熸垚",
      "type": "ai",
      "description": "Agent灏嗛€愬瓧绋挎媶鍒嗕负鍒嗛暅JSON鍜孲RT瀛楀箷",
      "aiConfig": {
        "systemPrompt": "你是专业的视频分镜脚本编写专家。请将逐字稿拆分为分镜段，每段约 40-45 个中文字（约 15 秒语速）。为每段生成英文视频提示词（prompt），描述画面内容、镜头类型、氛围和画质。",
        "userPromptTemplate": "请将以下逐字稿拆分为分镜脚本，每段约 ${segmentDuration} 秒。\n\n逐字稿：\n${transcript}\n\n请输出 JSON 数组，每项包含 id、act、narration、type、prompt、duration 字段，并同时输出 SRT 字幕文件内容。",
        "outputFormat": {
          "storyboard": "Array<{id, act, narration, type, prompt, duration}>",
          "srt": "String (SRT format subtitle content)"
        },
        "temperature": 0.3
      },
      "output": {
        "storyboard": "鍒嗛暅JSON鏁扮粍",
        "srt": "SRT瀛楀箷鏂囦欢鍐呭"
      }
    },
    {
      "step": 2,
      "name": "鎵归噺鐢熸垚瑙嗛绱犳潗",
      "type": "api",
      "description": "瀵规瘡涓垎闀滄骞惰璋冪敤鍗虫ⅵ瑙嗛鐢熸垚API",
      "forEach": "storyboard",
      "forEachAs": "segment",
      "parallel": {
        "maxConcurrent": 5,
        "delayMs": 1000
      },
      "branches": {
        "text2video": {
          "condition": "videoMethod === 'text2video'",
          "steps": [
            {
              "id": "2a",
              "name": "鏂囩敓瑙嗛",
              "skill": "jimeng/jimeng-video-v3-720p",
              "skillVariants": {
                "720p": "jimeng/jimeng-video-v3-720p",
                "1080p": "jimeng/jimeng-video-v3-1080p",
                "pro": "jimeng/jimeng-video-v3-pro"
              },
              "input": {
                "prompt": "${segment.prompt}",
                "method": "1",
                "frames": "${frames}",
                "config": {
                  "aspect_ratio": "${aspectRatio}"
                }
              },
              "output": "workIds[${segment.id}]"
            }
          ]
        },
        "img2video": {
          "condition": "videoMethod === 'img2video'",
          "steps": [
            {
              "id": "2b-1",
              "name": "文生图（首帧）",
              "skill": "jimeng/jimeng-img-v4",
              "input": {
                "prompt": "${segment.prompt}",
                "sizeDate": {
                  "width": "${aspectRatio === '16:9' ? 2560 : aspectRatio === '9:16' ? 1440 : 2048}",
                  "height": "${aspectRatio === '16:9' ? 1440 : aspectRatio === '9:16' ? 2560 : 2048}"
                }
              },
              "output": "imageWorkIds[${segment.id}]"
            },
            {
              "id": "2b-2",
              "name": "杞鍥剧墖缁撴灉",
              "skill": "jimeng/jimeng-task-query",
              "dependsOn": [
                "2b-1"
              ],
              "input": {
                "workId": "${imageWorkIds[segment.id]}",
                "routerName": "getImgV4"
              },
              "pollingConfig": {
                "intervalMs": 3000,
                "maxAttempts": 30,
                "completionCondition": "data.isFinish === true"
              }
            },
            {
              "id": "2b-3",
              "name": "鑾峰彇鍥剧墖URL",
              "skill": "jimeng/jimeng-work-result",
              "dependsOn": [
                "2b-2"
              ],
              "input": {
                "workId": "${imageWorkIds[segment.id]}"
              },
              "output": "imageUrls[${segment.id}]",
              "responseExtract": "images[0]"
            },
            {
              "id": "2b-4",
              "name": "鍥剧敓瑙嗛锛堥甯э級",
              "skill": "jimeng/jimeng-video-v3-720p",
              "dependsOn": [
                "2b-3"
              ],
              "input": {
                "prompt": "${segment.prompt}",
                "method": "2",
                "frames": "${frames}",
                "config": {
                  "image_urls": [
                    "${imageUrls[segment.id]}"
                  ]
                }
              },
              "output": "workIds[${segment.id}]"
            }
          ]
        }
      }
    },
    {
      "step": 3,
      "name": "杞瑙嗛鐢熸垚缁撴灉",
      "type": "api",
      "forEach": "workIds",
      "forEachAs": "workId",
      "parallel": {
        "maxConcurrent": 10,
        "delayMs": 500
      },
      "steps": [
        {
          "id": "3a",
          "name": "轮询任务状态",
          "skill": "jimeng/jimeng-task-query",
          "input": {
            "workId": "${workId}",
            "routerName": "${videoModel === '720p' ? 'getVideoV3_720p' : videoModel === '1080p' ? 'getVideoV3_1080p' : 'getVideoV3_Pro'}"
          },
          "pollingConfig": {
            "intervalMs": 5000,
            "maxAttempts": 60,
            "completionCondition": "data.isFinish === true",
            "timeoutMs": 300000
          }
        },
        {
          "id": "3b",
          "name": "鑾峰彇瑙嗛URL",
          "skill": "jimeng/jimeng-work-result",
          "dependsOn": [
            "3a"
          ],
          "input": {
            "workId": "${workId}"
          },
          "output": "videoResults[${segmentId}]",
          "responseExtract": {
            "videoUrl": "videos[0]",
            "images": "images"
          }
        }
      ]
    },
    {
      "step": 4,
      "name": "结果汇总",
      "type": "compute",
      "logic": "aggregateResults(storyboard, srt, videoResults)",
      "algorithm": {
        "merge": "灏嗗垎闀淛SON涓庤棰慤RL鍚堝苟",
        "stats": "缁熻鎴愬姛/澶辫触鏁伴噺",
        "failedRetry": "澶辫触鐨勫垎闀滄鏍囪涓篺ailedIds锛屽彲閲嶈瘯"
      },
      "output": "finalOutput"
    },
    {
      "step": 5,
      "name": "可选：一键成片",
      "type": "api",
      "optional": true,
      "condition": "user requests quick compilation && successCount <= 50 && totalDuration <= 30",
      "skill": "video-creation/quickly-video-create",
      "input": {
        "material_list": "${videoResults.map(r => ({ type: 'video', value: r.videoUrl }))}",
        "tags": "${extractKeywords(transcript)}",
        "proportion": "${aspectRatio === '9:16' ? '9:16' : aspectRatio === '1:1' ? '1:1' : '9:16'}",
        "video_duration": {
          "min": 15,
          "max": 30
        },
        "pre_id": "${Date.now().toString()}",
        "ai_voice": 1,
        "ai_bgm": 1,
        "ai_subtitle": 1
      },
      "output": "quicklyTaskId",
      "downstream": "video-creation/quickly-video-query"
    }
  ],
  "response": {
    "type": "object",
    "properties": {
      "storyboard": {
        "type": "array",
        "description": "鍒嗛暅鑴氭湰JSON",
        "items": {
          "type": "object",
          "properties": {
            "id": {
              "type": "string"
            },
            "act": {
              "type": "string"
            },
            "narration": {
              "type": "string"
            },
            "prompt": {
              "type": "string"
            },
            "duration": {
              "type": "string"
            },
            "videoUrl": {
              "type": "string",
              "description": "鐢熸垚鐨勮棰慤RL锛堟案涔呮湁鏁堬級"
            },
            "imageUrl": {
              "type": "string",
              "description": "生成的首帧图片 URL（img2video 模式）"
            }
          }
        }
      },
      "srt": {
        "type": "string",
        "description": "SRT瀛楀箷鏂囦欢鍐呭"
      },
      "segments": {
        "type": "array",
        "description": "视频片段汇总",
        "items": {
          "type": "object",
          "properties": {
            "id": {
              "type": "string"
            },
            "videoUrl": {
              "type": "string"
            },
            "duration": {
              "type": "number"
            },
            "prompt": {
              "type": "string"
            }
          }
        }
      },
      "totalSegments": {
        "type": "integer"
      },
      "successCount": {
        "type": "integer"
      },
      "failedCount": {
        "type": "integer"
      },
      "failedIds": {
        "type": "array",
        "items": {
          "type": "string"
        }
      }
    }
  },
  "relatedSkills": [
    "jimeng/jimeng-video-v3-720p",
    "jimeng/jimeng-video-v3-1080p",
    "jimeng/jimeng-video-v3-pro",
    "jimeng/jimeng-img-v4",
    "jimeng/jimeng-task-query",
    "jimeng/jimeng-work-result",
    "video-creation/quickly-video-create",
    "video-creation/quickly-video-query"
  ],
  "pricing": {
    "text2video_720p_5s": {
      "perSegment": 1.4,
      "unit": "CNY",
      "formula": "0.28元/秒 × 5秒"
    },
    "text2video_720p_10s": {
      "perSegment": 2.8,
      "unit": "CNY",
      "formula": "0.28元/秒 × 10秒"
    },
    "img2video_720p_5s": {
      "perSegment": 1.62,
      "unit": "CNY",
      "formula": "0.22鍏?鍥? + 0.28鍏?绉捗?绉?瑙嗛)"
    },
    "example_71segments_5s": {
      "text2video": "约99元",
      "img2video": "约115元"
    }
  },
  "limitations": {
    "videoDuration": "仅支持 5 秒（121 帧）或 10 秒（241 帧），无法精确匹配 SRT 时间戳",
    "noTTS": "鏃燭TS閰嶉煶鑳藉姏锛岄渶鐢ㄦ埛鍦ㄥ壀鏄犱腑閰嶉煶",
    "noAutoStitch": "暂无自动拼接长视频能力，quickly 最长 30 秒",
    "parallelLimit": "即梦 API 建议并行数不超过 5，避免触发频率限制"
  },
  "futureExtensions": [
    {
      "name": "tts-voiceover",
      "description": "TTS 配音 Skill：按逐字稿生成语音音频",
      "status": "planned"
    },
    {
      "name": "video-stitch",
      "description": "视频拼接 Skill：FFmpeg 服务端按 SRT 时间轴精确拼接",
      "status": "planned"
    },
    {
      "name": "seeddance-video",
      "description": "seeddance2.0 模型接入：生成更高质量视频",
      "status": "planned"
    }
  ],
  "tokenConfig": {
    "type": "bearer",
    "configFile": "~/.openclaw/voc-credentials.json",
    "tokenField": "vocToken",
    "resolutionOrder": [
      "configFile"
    ],
    "apigId": "Vo3ROWEvDy",
    "paymentUrlResolution": {
      "description": "动态构建充值URL：在已存在 session token 时查出用户 objectId，拼接到专属充值页面",
      "steps": [
        "1. 从 voc-credentials.json 获取当前 session token",
        "2. 调用 GET https://server.fmode.cn/parse/users/me （Header: X-Parse-Application-Id: ncloudmaster, X-Parse-Session-Token: {token}）获取用户 objectId",
        "3. 拼接充值URL: https://app.fmode.cn/dev/apig-pay/?user={objectId}&apigid=Vo3ROWEvDy&fun_id=HOkkX72PMF",
        "4. 将该URL展示给用户，引导扫码支付"
      ],
      "userResolveEndpoint": "https://server.fmode.cn/parse/users/me",
      "userResolveHeaders": {
        "X-Parse-Application-Id": "ncloudmaster",
        "X-Parse-Session-Token": "{vocToken}"
      },
      "paymentBaseUrl": "https://app.fmode.cn/dev/apig-pay/",
      "paymentParams": {
        "user": "{resolvedUserId}",
        "apigid": "Vo3ROWEvDy",
        "fun_id": "HOkkX72PMF"
      },
      "balanceCheckEndpoint": "https://server.fmode.cn/api/apig/getApig"
    },
    "onMissing": {
      "action": "showPaymentQR",
      "qrCodeUrl": "https://app.fmode.cn/dev/apig-pay/?apigid=Vo3ROWEvDy&fun_id=HOkkX72PMF",
      "title": "扫码开通 VOC-AI 数据服务",
      "message": "此 Skill 需要有效的 API Token。请先打开充值页面完成登录和充值；当用户提供 session token 后，请由 OpenClaw 运行 node scripts/tools/set-voc-token.js <session-token> 写入 ~/.openclaw/voc-credentials.json 的 vocToken 字段，再重试当前 Skill。",
      "tokenSetupCommandTemplate": "node scripts/tools/set-voc-token.js {sessionToken}",
      "tokenSetupInstructions": "当用户完成登录和充值后，如果用户把 session token 发给你，请直接运行 node scripts/tools/set-voc-token.js <session-token> 写入 ~/.openclaw/voc-credentials.json，然后重试当前 Skill。"
    },
    "onBalanceInsufficient": {
      "action": "resolveUserThenShowPayment",
      "title": "VOC-AI Token 余额不足，请扫码充值",
      "message": "当前 Token 余额不足，我将为你生成专属充值链接，扫码支付后即可继续使用。"
    }
  },
  "errorHandling": {
    "balanceInsufficient": {
      "conditions": [
        {
          "responseField": "code",
          "operator": "in",
          "value": [
            -2,
            -3,
            -10,
            402,
            429
          ]
        },
        {
          "responseField": "msg",
          "operator": "contains",
          "value": [
            "余额不足",
            "insufficient",
            "balance",
            "quota",
            "没有开通",
            "权限或余额"
          ]
        },
        {
          "responseField": "mess",
          "operator": "contains",
          "value": [
            "余额不足",
            "insufficient",
            "balance",
            "quota",
            "没有开通",
            "权限或余额"
          ]
        },
        {
          "responseField": "message",
          "operator": "contains",
          "value": [
            "余额不足",
            "insufficient",
            "balance",
            "没有开通",
            "权限或余额"
          ]
        }
      ],
      "matchMode": "any",
      "trigger": "tokenConfig.onBalanceInsufficient"
    },
    "unauthorized": {
      "conditions": [
        {
          "responseField": "code",
          "operator": "in",
          "value": [
            401,
            403
          ]
        },
        {
          "responseField": "msg",
          "operator": "contains",
          "value": [
            "unauthorized",
            "token",
            "invalid",
            "auth"
          ]
        }
      ],
      "matchMode": "any",
      "trigger": "tokenConfig.onMissing"
    }
  },
  "timeout": 600000,
  "retry": {
    "maxAttempts": 2,
    "delay": 3000,
    "backoffMultiplier": 2
  }
}