Audio -> Text dataset. Provides well aligned audio utterances. Two special tokens are used. Text format: Begin episode

Collates data

Structure that contains necessary states for beam search.

r""" https://github.com/cybertronai/pytorch-lamb Implements Lamb algorithm. It has been proposed in `Large Batch Optimization for Deep Lea

Performs log melspec transformation on raw audio Adapted from https://github.com/zcaceres/spec_augment/blob/master/SpecAugment.ipynb P

Modified version of decoder layer that caches the attention weights for analysis

Collates data

Audio -> Text dataset. Provides well aligned audio utterances. Two special tokens are used. Text format: Begin episode <EOS>

Acoustic model

Audio -> Text dataset. During training we obtain a random segment of audio and ask the model to predict all utterances within that c

The core Recurent Neural Network used by UIS-RNN.

Represents a "frame" of audio data.

Represents a "frame" of audio data.

r"""Inject some information about the relative or absolute position of the tokens in the sequence. The positional encodings have the same dim

Collates data

Fairseq implementation Implements Adafactor algorithm. https://github.com/pytorch/fairseq/blob/master/fairseq/optim/adafactor.py This

Collates data

A class for printing logging information to screen.

A dataset that produces random audio segments from a directory

Unbounded Interleaved-State Recurrent Neural Networks.