Loader

Defines the indexer class and the loader class, handling batching of the dataset to train the network.

Indexer

Handles returning the next indices of the batch to load into memory, to train the network with.

indexer:__init(_dir, batchSize)

dirPath Directory containing the LMDB data folders for spectrogram, labels and transcripts.

batchSize The sizes of each batch to create.

indexer:prep_sorted_inds()

Sorts the pointers of each training sample by length, and stores the order in the class for retrieving batches.

indexer:nxt_sorted_inds()

Returns the next set of sorted indices that can be loaded from disk based on the batch size (iterator).

indexer:nxt_same_len_inds()

Returns the next set of indices that can be loaded from disk that have the same length (iterator).

indexer:nxt_inds()

Returns the next set of indices based purely on batch size (no ordering).

Loader

Loads batches of data from LMDB files used in training/testing.

Loader:__init(dirPath)

dirPath Directory containing the LMDB data folders for spectrogram, labels and transcripts.

Loader:nxt_batch(indices, includeTranscripts)

Returns the next batch of the dataset based on the given indices.

indices The indices of the test samples that need to be retrieved. This is handled by the Indexer class above.

includeTranscripts Set to true if transcripts are needed as well for each training sample.