API Reference¶
Below is a diagram of the main classes used in MFA:
Aligner API¶
There are two main Aligner classes, one for using a pretrained model and one for training a model while aligning. A class diagram of the Aligner API can be found below:
PretrainedAligner(corpus, dictionary, …[, …]) |
Class for aligning a dataset using a pretrained acoustic model |
TrainableAligner(corpus, dictionary, …[, …]) |
Aligner that aligns and trains acoustics models on a large dataset |
Corpus API¶
The Corpus class contains information about how a dataset is structured. A class diagram of the Corpus API can be found below:
Corpus(directory, output_directory[, …]) |
Class that stores information about the dataset to align. |
Dictionary API¶
The Dictionary class contains pronunciation and orthographic information. A class diagram of the Dictionary API can be found below:
Dictionary(input_path, output_directory[, …]) |
Class containing information about a pronunciation dictionary |
Model API¶
Output from training a Model is compressed using the Archive class, which results in a zip folder. A class diagram of the Model API can be found below:
AcousticModel(source[, is_tmpdir]) |
|
G2PModel(source[, is_tmpdir]) |
|
IvectorExtractor(source[, is_tmpdir]) |
Archive for i-vector extractors (used with DNNs) |
Feature processing API¶
mfcc(mfcc_directory, num_jobs, …) |
Multiprocessing function that converts wav files into MFCCs |
apply_cmvn(directory, num_jobs, config) |
|
add_deltas(directory, num_jobs, config) |
|
apply_lda(directory, num_jobs, config) |
Multiprocessing API¶
The multiprocessing module contains most of the interactions with Kaldi, as multiple processes are used to speed up the set up and aligning of the dataset.
compile_train_graphs(directory, …[, debug]) |
Multiprocessing function that compiles training graphs for utterances |
mono_align_equal(mono_directory, …) |
Multiprocessing function that creates equal alignments for base monophone training |
align(iteration, directory, split_directory, …) |
Multiprocessing function that aligns based on the current model |
acc_stats(iteration, directory, …) |
Multiprocessing function that computes stats for GMM training |
tree_stats(directory, align_directory, …) |
Multiprocessing function that computes stats for decision tree training |
calc_fmllr(directory, split_directory, …) |
Multiprocessing function that computes speaker adaptation (fMLLR) |
convert_alignments(directory, …) |
Multiprocessing function that converts alignments from previous training |
convert_ali_to_textgrids(align_config, …) |
Multiprocessing function that aligns based on the current model |
For use with DNNs¶
lda_acc_stats(directory, split_dir, …) |
Multiprocessing function that accumulates LDA statistics |
calc_lda_mllt(directory, split_directory, …) |
Multiprocessing function that calculates LDA+MLLT transformations |
gmm_gselect(config, num_jobs) |
Multiprocessing function that stores Gaussian selection indices on disk |
acc_global_stats(config, num_jobs, iteration) |
Multiprocessing function that accumulates global GMM stats |
gauss_to_post(config, num_jobs) |
Multiprocessing function that does Gaussian selection and posterior extraction |
acc_ivector_stats(config, num_jobs, iteration) |
Multiprocessing function that calculates i-vector extractor stats |
extract_ivectors(config, num_jobs) |
Multiprocessing function that extracts i-vectors. |
get_egs(config, ali_dir, valid_uttlist, …) |
Multiprocessing function that gets training examples for the neural net |
get_lda_nnet(config, align_directory, num_jobs) |
Multiprocessing function that extracts training examples and does LDA transformation |
nnet_train_trans(nnet_dir, align_dir, …) |
Multiprocessing function that trains transition prbabilities and sets priors. |
nnet_train(nnet_dir, egs_dir, mdl, i, num_jobs) |
Multiprocessing function that trains the neural net. |
nnet_align(i, config, train_directory, …) |
Multiprocessing function that generates an nnet alignment |
compute_prob(i, nnet_dir, egs_dir, …) |
Multiprocessing function that computes the current log probability of the iteration |
get_average_posteriors(i, nnet_dir, …) |
Multiprocessing function that gets average posterior for purposes of computing priors (for nnet) |
relabel_egs(i, nnet_dir, egs_in, alignments, …) |
Multiprocessing function that relabels training examples |
Trainer API¶
These Trainer classes contain information about configuring data preparation and training. A class diagram of the Configuration API can be found below:
MonophoneTrainer(default_feature_config) |
Configuration class for monophone training |
TriphoneTrainer(default_feature_config) |
Configuration class for triphone training |
LdaTrainer(default_feature_config) |
Configuration class for LDA+MLLT training |
SatTrainer(default_feature_config) |
Configuration class for speaker adapted training (SAT) |
IvectorExtractorTrainer(default_feature_config) |
Configuration class for i-vector extractor training |
NnetTrainer(default_feature_config) |
Configuration class for neural network training |