athena.data.feature_normalizer¶
Feature Normalizer
Module Contents¶
Classes¶
FeatureNormalizer |
Feature Normalizer |
Functions¶
compute_cmvn_by_chunk_for_all_speaker(feature_dim, speakers, featurizer, entries) |
computing mean and variance for all speakers in a multi-process way |
-
athena.data.feature_normalizer.compute_cmvn_by_chunk_for_all_speaker(feature_dim, speakers, featurizer, entries)¶ computing mean and variance for all speakers in a multi-process way
-
class
athena.data.feature_normalizer.FeatureNormalizer(cmvn_file=None)¶ Feature Normalizer
-
__call__(self, feat_date, speaker, reverse=False)¶
-
apply_cmvn(self, feat_data, speaker, reverse=False)¶ TODO: docstring
-
compute_cmvn(self, entries, speakers, featurizer, feature_dim, num_cmvn_workers=1)¶ Compute cmvn for filtered entries
-
compute_cmvn_kaldiio(self, entries, speakers, kaldi_io_feats, feature_dim)¶ Compute cmvn for filtered entries using kaldi-format data
-
load_cmvn(self)¶ TODO: docstring
-
save_cmvn(self)¶ TODO: docstring
-