数字人口播

本文档介绍如何在 aijisu 中使用 Kling Avatar v2 异步数字人视频生成接口。

当前支持模型：

模型名称	类型	适合场景	计费方式
kling-avatar-v2-standard	数字人视频	常规口播、客服讲解、课程导语、营销短视频	按输出视频秒数计费
kling-avatar-v2-pro	数字人视频	更高质量数字人口播、正式宣传片、品牌视频、重要课程内容	按输出视频秒数计费

接口采用异步任务模式：

操作	方法	路径
提交数字人视频任务	POST	/v1/video/generations
查询数字人视频任务	GET	/v1/video/generations/{task_id}

---

1. 通用鉴权

所有接口都需要在请求头中携带 API Key：

Authorization: Bearer sk-xxxxxxxxxxxxxxxx
Content-Type: application/json

示例域名：

https://api.xxx.xx

---

2. 模型简介

Kling Avatar v2 用于根据人物图片和音频生成数字人口播视频。用户需要提供一张人物图片和一段音频，模型会生成带口型和面部动作的视频。

支持两个版本：

模型	说明
kling-avatar-v2-standard	标准版本，适合常规口播和批量生成
kling-avatar-v2-pro	Pro 版本，适合对画面、表情、稳定性要求更高的正式内容

---

3. 提交数字人视频任务

POST https://api.xxx.xx/v1/video/generations

3.1 请求参数

参数	类型	必填	说明
model	string	是	kling-avatar-v2-standard 或 kling-avatar-v2-pro
image_url	string	是	人物图片 URL
audio_url	string	是	音频 URL，用于驱动口型
audio_duration_seconds	number	是	音频秒数，用于提交前预扣费估算
prompt	string	否	风格提示，例如场景、人物状态、镜头感觉

注意：

• audio_duration_seconds 必须传。
• 该字段用于提交前估价，避免无法预扣费。
• 请尽量传入音频真实秒数，避免费用预估和实际任务时长差异过大。

3.2 提交响应示例

{
  "id": "task_xxxxxxxxxxxxx",
  "task_id": "task_xxxxxxxxxxxxx",
  "object": "video.generation.job",
  "status": "queued"
}

字段说明：

字段	说明
id / task_id	异步任务 ID，用于查询结果
status	任务状态，常见值为 queued、processing、completed、failed

---

4. 查询数字人视频任务

GET https://api.xxx.xx/v1/video/generations/{task_id}

查询示例：

curl -X GET "https://api.xxx.xx/v1/video/generations/task_xxxxxxxxxxxxx" \
  -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx"

完成响应示例：

{
  "id": "task_xxxxxxxxxxxxx",
  "task_id": "task_xxxxxxxxxxxxx",
  "object": "video.generation.job",
  "status": "completed",
  "video_url": "https://api.xxx.xx/media/xxxxx.mp4",
  "result": {
    "outputs": [
      "https://api.xxx.xx/media/xxxxx.mp4"
    ]
  }
}

---

5. 请求示例

5.1 Standard 生成客服讲解视频

curl -X POST "https://api.xxx.xx/v1/video/generations" \
  -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-avatar-v2-standard",
    "image_url": "https://cdn.example.com/avatar/customer-service.png",
    "audio_url": "https://cdn.example.com/audio/customer-intro.mp3",
    "audio_duration_seconds": 12,
    "prompt": "friendly customer service presenter, natural expression, clean studio lighting"
  }'

5.2 Standard 生成课程导语

curl -X POST "https://api.xxx.xx/v1/video/generations" \
  -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-avatar-v2-standard",
    "image_url": "https://cdn.example.com/avatar/teacher.png",
    "audio_url": "https://cdn.example.com/audio/course-opening.mp3",
    "audio_duration_seconds": 18.5,
    "prompt": "professional teacher, warm smile, stable camera, natural lip sync"
  }'

5.3 Pro 生成品牌口播视频

curl -X POST "https://api.xxx.xx/v1/video/generations" \
  -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-avatar-v2-pro",
    "image_url": "https://cdn.example.com/avatar/brand-host.png",
    "audio_url": "https://cdn.example.com/audio/brand-message.mp3",
    "audio_duration_seconds": 24,
    "prompt": "premium corporate spokesperson video, clean studio background, confident expression"
  }'

5.4 Pro 生成新闻播报

curl -X POST "https://api.xxx.xx/v1/video/generations" \
  -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-avatar-v2-pro",
    "image_url": "https://cdn.example.com/avatar/news-anchor.png",
    "audio_url": "https://cdn.example.com/audio/news-briefing.mp3",
    "audio_duration_seconds": 32.7,
    "prompt": "news anchor style, formal tone, stable frontal shot, neutral expression"
  }'

---

6. 常见错误

6.1 缺少 image_url

数字人视频必须提供人物图片：

{
  "image_url": "https://cdn.example.com/avatar.png"
}

6.2 缺少 audio_url

数字人视频必须提供驱动口型的音频：

{
  "audio_url": "https://cdn.example.com/voice.mp3"
}

6.3 缺少 audio_duration_seconds

错误请求：

{
  "model": "kling-avatar-v2-standard",
  "image_url": "https://cdn.example.com/avatar.png",
  "audio_url": "https://cdn.example.com/voice.mp3"
}

修复方式：补充音频秒数：

{
  "audio_duration_seconds": 12.5
}

---

7. 推荐工作流

1. 准备人物图片。
2. 使用 TTS 或录音生成一段音频。
3. 获取音频实际秒数。
4. 提交 /v1/video/generations，传入 image_url、audio_url、audio_duration_seconds。
5. 轮询 /v1/video/generations/{task_id}。
6. 任务完成后读取 video_url。

---

8. 最小可用请求

8.1 Avatar Standard

curl -X POST "https://api.xxx.xx/v1/video/generations" \
  -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-avatar-v2-standard",
    "image_url": "https://cdn.example.com/avatar.png",
    "audio_url": "https://cdn.example.com/voice.mp3",
    "audio_duration_seconds": 10
  }'

8.2 Avatar Pro

curl -X POST "https://api.xxx.xx/v1/video/generations" \
  -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-avatar-v2-pro",
    "image_url": "https://cdn.example.com/avatar.png",
    "audio_url": "https://cdn.example.com/voice.mp3",
    "audio_duration_seconds": 10
  }'