字幕同步：把文字稿对齐到音频时间轴

{
  "type": "object",
  "required": [
    "body"
  ],
  "properties": {
    "body": {
      "type": "object",
      "$schema": "https://json-schema.org/draft/2020-12/schema",
      "required": [
        "audio_url",
        "text"
      ],
      "~standard": {
        "vendor": "zod",
        "version": 1,
        "jsonSchema": {}
      },
      "properties": {
        "text": {
          "type": "string",
          "maxLength": 675000,
          "minLength": 1,
          "description": "要对齐的文字，必须是纯字符串，不能包在 JSON 对象中，也不能包含说话人标注。"
        },
        "audio_url": {
          "type": "string",
          "format": "uri",
          "description": "音频文件的 HTTPS 地址，可用公开或预签名链接，文件最大 20MB。Monid 会在服务端获取文件，再转发给 ElevenLabs。"
        },
        "expected_duration_minutes": {
          "type": "number",
          "maximum": 600,
          "description": "仅供 Monid 计费使用，不发送给上游：预计音频时长，单位为分钟，用于确定调用前的预留额度。最终按实际对齐的时长结算；省略时按 60 分钟预留。",
          "exclusiveMinimum": 0
        }
      },
      "description": "将转写文字与音频对齐，生成逐词时间戳。",
      "additionalProperties": false
    }
  },
  "additionalProperties": false
}