Writing a Voice Control Wrapper

To implement a wrapper for a new voice control system, write a class definition that extends VoiceControlWrapper. This class should implement an extract_embedding method to extract speaker embeddings, and an inference method to perform inference. A template appears below:

class MyVoiceControlWrapper(VoiceControlWrapper):
    def __init__(self):
        # Put code here to load model checkpoints and initialize the VC system

    def extract_embedding(self, source_file: str) -> torch.Tensor:
        """Extract the speaker embedding from a source .wav file"""
        # Put code here to extract a single speaker embedding from a source file

    def inference(self, source_file: str, output_file: str,
                  source_embedding: torch.Tensor, target_embedding: torch.Tensor):
        """Perform inference with a source file and target speaker embedding,
        writing to the output file"""
        # Put code here to perform inference

After defining this class, an instance of it can be used to construct an Anonymizer and perform differentially private anonymization in the same way as previous examples.