athena.data.feature_normalizer

Feature Normalizer

Module Contents

Classes

FeatureNormalizer	Feature Normalizer
FS2FeatureNormalizer	Fastspeech2 Feature Normalizer

class athena.data.feature_normalizer.FeatureNormalizer(cmvn_file=None)

Feature Normalizer

__call__(feat_data, speaker, reverse=False)

apply_cmvn(feat_data, speaker, reverse=False)

transform original feature to normalized feature

compute_cmvn(entries, speakers, featurizer, feature_dim, num_cmvn_workers=1)

compute cmvn for filtered entries

compute_cmvn_by_chunk_for_all_speaker(feature_dim, speakers, featurizer, entries)

because of memory issue, we used incremental approximation for the calculation of cmvn

compute_cmvn_kaldiio(entries, speakers, kaldi_io_feats, feature_dim)

compute cmvn for filtered entries using kaldi-format data

load_cmvn()

load mean and var

save_cmvn(variable_list)

save cmvn variables determined by variable_list to file

Parameters

variable_list (list) – e.g. [“speaker”, “mean”, “var”]

class athena.data.feature_normalizer.FS2FeatureNormalizer(cmvn_file=None)

Bases: FeatureNormalizer

Fastspeech2 Feature Normalizer

__call__(feat_data, speaker, feature_type='mel', reverse=False)

compute_fs2_cmvn(entries, speakers, num_cmvn_workers=1)

compuate cmvn of mel-spec,f0 and energy

apply_cmvn(feat_data, speaker, feature_type='mel', reverse=False)

transform original feature to normalized feature

load_cmvn()

load mel_mean, mel_var, f0_mean, f0_var and energy_mean, energy_var