Builds and dumps variables and a manifest file.

Decodes audio file paths into 32-bit floating point vectors. Args: fp: Audio file path. fs: If specified, resamples decoded audio to this r

Creates a manifest file with given name and dumps meta information related to model. Parameters ---------- filename : str, optional

Creates a file with given name and dumps byte weights in it. Parameters ---------- variable_name : str Name of given variable

Returns Checkpoint dumper instance for a given model type. Parameters ---------- model_type : str Type of deeplearning framework checkp

Makes directory if not existing. Parameters ---------- directory : str Path to directory

Checks whether name should be ignored or not. Parameters ---------- name : str Name to be checked Returns ------- bo

Converts variable names to standard file names. Parameters ---------- var_name : str Variable name to be converted Returns

Constructs object for Checkpoint Dumper. Parameters ---------- checkpoint_file : str Path to the model checkpoint output_dir

Constructs object for Tensorflow Checkpoint Dumper. Parameters ---------- checkpoint_file : str Path to the model checkpoint

Decodes audio file paths into mini-batches of samples. Args: fps: List of audio file paths. batch_size: Number of items in the batch. s