x : torch.Tensor, shape = (batch_size, n_mels, n_ctx) the mel spectrogram of the audio

Generate batch from context tokens.

Converts a list of labels into a binary vector Args: label: list of integer labels num_labels: total number of labels Re

Returns a dictionary of the label vocabulary mapping the label column to the idx column. key sets whether the label or idx is the key in the

Save only the vocabulary of the tokenizer (vocabulary). Returns: `Tuple(str)`: Paths to the files saved.

This function has been mostly taken from huggingface conversational ai code at https://medium.com/huggingface/how-to-build-a-state-of-the-

Pad or trim the audio array to N_SAMPLES, as expected by the encoder.