Instructions

Implementation Reference

• src/mcp_server_baidu_digitalhuman/dhserver.py:258-316 (handler)

  The handler function that executes the generateDhVideo tool. It constructs a VideoGenerateRequest with provided parameters and calls the DH API client to generate the video asynchronously.

  async def generateDhVideo(
          figureId: Annotated[str, Field(description="人像ID", default=None)],
          voiceId: Annotated[str, Field(description="音色ID", default=None)],
          text: Annotated[str, Field(description="播报内容", default=None)],
          inputAudioUrl: Annotated[str, Field(description="驱动音频URL", default=None)],
          resolutionWidth: Annotated[int, Field(description="分辨率:宽", default=768)],
          resolutionHeight: Annotated[int, Field(description="分辨率:高", default=1280)],
          backgroundTransparent: Annotated[bool, Field(description="背景是否透明", default=False)],
          cameraId: Annotated[int,
              Field(description="数字人相机机位，0:横屏半身, 1:竖屏半身, 2: 横屏全身, 3: 竖屏全身", default=3)],
          backgroundImageUrl: Annotated[str, Field(description="背景图片", default=None)],
          callbackUrl: Annotated[str, Field(description="回调地址", default=None)],
          driveType: Annotated[Literal["TEXT", "VOICE"],
              Field(description="驱动类型, TEXT:文本驱动, VOICE: 音频驱动", default="TEXT")],
          subtitleEnable: Annotated[bool, Field(description="是否启用字幕", default=False)],
          autoAnimoji: Annotated[bool, Field(description="自动添加数字人动作", default=False)]
  ) -> MCPVideoGenerateResponse:
      """
      Generate a new digital human video using the DH API.
  
      Args:
          figureId: 人像ID
          driveType: 驱动类型, TEXT:文本驱动, VOICE: 音频驱动
          text: 文本内容，播报内容
          voiceId: 音色id,
          inputAudioUrl: 驱动音频URL
          resolutionWidth: 分辨率宽
          resolutionHeight: 分辨率高
          backgroundTransparent: 背景透明
          cameraId: 0:横屏半身, 1:竖屏半身, 2: 横屏全身, 3: 竖屏全身
          subtitleEnable: 字幕
          backgroundImageUrl: 背景图片
          autoAnimoji: 自动添加数字人动作
          callbackUrl: 回调地址
  
      Returns:
          taskId: 任务ID
      """
      try:
          request = VideoGenerateRequest(
              figureId=figureId,
              driveType=driveType,
              text=text,
              ttsParams=TtsParams(person=str(voiceId), speed="5", volume="5", pitch="5"),
              inputAudioUrl=inputAudioUrl,
              videoParams=VideoParams(width=resolutionWidth, height=resolutionHeight, transparent=backgroundTransparent),
              dhParams=DHParams(cameraId=cameraId),
              subtitleParams=SubtitleParams(subtitlePolicy="SRT", enabled=True) if subtitleEnable else None,
              backgroundImageUrl=backgroundImageUrl,
              callbackUrl=callbackUrl,
              autoAnimoji=autoAnimoji,
          )
  
          client = await getDhClient()
          ret = await client.generate_avatar_video(request)
          return ret
      except Exception as e:
          return MCPVideoGenerateResponse(error=str(e))

• src/mcp_server_baidu_digitalhuman/dhserver.py:239-257 (registration)

  The @mcp.tool decorator registers the generateDhVideo tool with its name and detailed description including usage examples.

  @mcp.tool(
      name="generateDhVideo",
      description=(
      """
  #工具说明：根据所选数字人像ID及发音人ID，生成数字人视频。
  # 样例1：
  用户输入：用数字人像ID为xxx，发音人ID为yyy的音色，视频的内容是“大家好，我是数字人播报的内容”，使用横屏全身的机位，视频背景用\
  “https://digital-human-material.bj.bcebos.com/-%5BLjava.lang.String%3B%4046f6cc1e.png”，\
  开启自动添加动作，开启字幕，生成一个1080P的数字人视频。
  思考过程:
  1.用户想要用人像ID生成一个数字人视频，对声音，背景，字幕，分辨率等有要求，不是一个简单的数字人视频，需要使用“generateDhVideo”工具。
  2.工具需要FigureId，driveType，text，person，inputAudioUrl，width，hight，cameraID，enable，backgroundimageUrl，\
  autoAnimoji这些参数。
  3.FigureId是需要使用的人像ID，所以值为xxx。给的播报内容是文本，所以driveType是文本驱动，text为“大家好，我是数字人播报的内容”。\
  发音人已经提供了ID，所以person的值是yyy，开启自动动作，所以autoAnimoji的值为true，开启字幕，所以enabled的值为true，分辨率为1080P，\
  拆分为width的值为1920，hight的值为1080，backgroundimageUrl的值是\
  “https://digital-human-material.bj.bcebos.com/-%5BLjava.lang.String%3B%4046f6cc1e.png”
      """)
  )

• src/mcp_server_baidu_digitalhuman/dhserver.py:258-274 (schema)

  Input schema defined via Annotated types and Field descriptions in the function signature for MCP tool validation.

  async def generateDhVideo(
          figureId: Annotated[str, Field(description="人像ID", default=None)],
          voiceId: Annotated[str, Field(description="音色ID", default=None)],
          text: Annotated[str, Field(description="播报内容", default=None)],
          inputAudioUrl: Annotated[str, Field(description="驱动音频URL", default=None)],
          resolutionWidth: Annotated[int, Field(description="分辨率:宽", default=768)],
          resolutionHeight: Annotated[int, Field(description="分辨率:高", default=1280)],
          backgroundTransparent: Annotated[bool, Field(description="背景是否透明", default=False)],
          cameraId: Annotated[int,
              Field(description="数字人相机机位，0:横屏半身, 1:竖屏半身, 2: 横屏全身, 3: 竖屏全身", default=3)],
          backgroundImageUrl: Annotated[str, Field(description="背景图片", default=None)],
          callbackUrl: Annotated[str, Field(description="回调地址", default=None)],
          driveType: Annotated[Literal["TEXT", "VOICE"],
              Field(description="驱动类型, TEXT:文本驱动, VOICE: 音频驱动", default="TEXT")],
          subtitleEnable: Annotated[bool, Field(description="是否启用字幕", default=False)],
          autoAnimoji: Annotated[bool, Field(description="自动添加数字人动作", default=False)]
  ) -> MCPVideoGenerateResponse: