infer module

infer.predict(model, audio, device)

Function that return the prediction of the model, i.e the argmax of the softmax of the logits.

Parameters:

• model (nn.Module) – pytorch neural network

• audio (torch.tensor) – audio tensor

• device (torch.device) – Preferably cpu

Returns:

predicted label

Return type:

str

infer.record(seconds=1, sample_rate=16000)

Function to record the user using the machine default microphone

Parameters:

• seconds (int, optional) – Defaults to 1.

• sample_rate (int, optional) – Defaults to 16000.

Yields:

torch.tensor – A 1 dimensional tensor representing the audio