input params that will use to connect to speech server. params: productId,publicKey and secretKey are provide by SPEECH(思必驰)

在_init__中对模型文件进行加载，并在initdata 中构建输入输出函数句柄。 init tensorflow enviroment and load graph.

Articulation Layer Args: kernels_size: list, the kernel size of each conv2d layer E: int, the channels of the emotion layer

Audio2Face Model Args: output_size: int, the output size of the output layer keep_pro: float, the keep probability of the dropout

model: speech_recognition

Formant Layer Args: kernels_size: list, the kernel size of each conv2d layer outputs: list, the output channels of each conv2d lay

Output Layer Args: output_size: int, the output size of the output layer keep_pro: float, the keep probability of the dropout laye

Inference with tflite model Args: model_path: str, the path of the tflite model pb_model_path: str, the path of the pb model, if

init udp server, and bind to addr addr_bind.

init the send to address addr_send.