Open an audio file and read as mono waveform, resampling as necessary Parameters ---------- file: str The audio file to open

Converts a string in a sequence of tokens. Args: text (`str`): The sequence to be encoded. a

Write a dict or a list of dicts into a JSON file :param data: Data to write :type data: list[dict[str, any]] | dict[str, any] :param path

:param label_to_idx: Map from label string to integer index. :param name: Override the name of this scoring function. :param

Compute the score based on the predictions and targets. This is a private function and the metric should be used as a functor

Build masks and position id for left to right model.

Compute the log-Mel spectrogram of Parameters ---------- audio: Union[str, np.ndarray, torch.Tensor], shape = (*) The path t

load the mel filterbank matrix for projecting STFT into a Mel spectrogram. Allows decoupling librosa dependency; saved using: np.sav

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

Reformat annotations to the MSCOCO format :param predictions: List of predicted captions :type predictions: list[str] :param ground_truth

Remove punctuation based on Unicode category.

Returns sinusoids for positional embedding

Valid return types for the metric are - tuple(tuple(string: name of the subtype, float: the value)): This is the case with se

Calls the compute function of the metric, and after validating the output, returns the metric score