speechbrain.utils

Package containing various tools (accuracy, checkpoints …)

speechbrain.utils.Accuracy	Calculate accuracy.
speechbrain.utils.DER	Calculates Diarization Error Rate (DER) which is the sum of Missed Speaker (MS), False Alarm (FA), and Speaker Error Rate (SER) using md-eval-22.pl from NIST RT Evaluation.
speechbrain.utils.EDER	Calculates Emotion Diarization Error Rate (EDER) which is the sum of Missed Emotion (ME), False Alarm (FA), and Confusion (CF).
speechbrain.utils.autocast	This module implements utilities and abstractions for use with torch.autocast, i.e. Automatic Mixed Precision.
speechbrain.utils.bleu	Library for computing the BLEU score
speechbrain.utils.callchains	Chaining together callables, if some require relative lengths
speechbrain.utils.checkpoints	This module implements a checkpoint saver and loader.
speechbrain.utils.data_pipeline	A pipeline for data transformations.
speechbrain.utils.data_utils	This library gathers utilities for data io operation.
speechbrain.utils.depgraph	A dependency graph for finding evaluation order.
speechbrain.utils.distributed	Guard for running certain operations on main process only
speechbrain.utils.dynamic_chunk_training	Configuration and utility classes for classes for Dynamic Chunk Training, as often used for the training of streaming-capable models in speech recognition.
speechbrain.utils.edit_distance	Edit distance and WER computation.
speechbrain.utils.epoch_loop	Implements a checkpointable epoch counter (loop), optionally integrating early stopping.
speechbrain.utils.fetching	Downloads or otherwise fetches pretrained models
speechbrain.utils.filter_analysis	Implements utils to model and combine filter properties, i.e. compute how window size, stride, etc.
speechbrain.utils.hparams	Utilities for hparams files
speechbrain.utils.hpopt	Utilities for hyperparameter optimization.
speechbrain.utils.kmeans	Utilities for training kmeans model.
speechbrain.utils.logger	Managing the logger, utilities
speechbrain.utils.metric_stats	The metric_stats module provides an abstract class for storing statistics produced over the course of an experiment and summarizing them.
speechbrain.utils.optimizers	Implements functions to avoid optimizing certain parameters
speechbrain.utils.parallel	Parallel processing tools to help speed up certain tasks like data preprocessing.
speechbrain.utils.parameter_transfer	Convenience functions for the simplest parameter transfer cases.
speechbrain.utils.pretrained	Training utilities for pretrained models
speechbrain.utils.profiling	Wrapper to handle PyTorch profiling and benchmarking.
speechbrain.utils.streaming	Utilities to assist with designing and training streaming models.
speechbrain.utils.superpowers	Superpowers which should be sparingly used.
speechbrain.utils.text_to_sequence	from https://github.com/keithito/tacotron
speechbrain.utils.torch_audio_backend	Library for checking the torchaudio backend.
speechbrain.utils.train_logger	Loggers for experiment monitoring.

Summary

Reference