IvectorCorpusMixin

class montreal_forced_aligner.corpus.ivector_corpus.IvectorCorpusMixin(**kwargs)

Bases: AcousticCorpusMixin, IvectorConfigMixin

Abstract corpus mixin for corpora that extract ivectors

See also

AcousticCorpusMixin

For corpus parsing parameters

IvectorConfigMixin

For ivector extraction parameters

adapt_plda()

Adapted a trained PLDA model with new ivectors

property adapted_plda_path

Path to adapted PLDA model

collect_speaker_ivectors()

Collect trained per-speaker ivectors

collect_utterance_ivectors()

Collect trained per-utterance ivectors

compute_plda(minimum_utterance_count=1)

Train a PLDA model

compute_speaker_ivectors()

Calculated and save per-speaker ivectors as the mean over their utterances

property dubm_path

DUBM model path

extract_ivectors()

Multiprocessing function that extracts job_name-vectors.

See also

ExtractIvectorsFunction

Multiprocessing helper function for each job

IvectorCorpusMixin.extract_ivectors_arguments

Job method for generating arguments for helper function

sid/extract_ivectors.sh

Reference Kaldi script

extract_ivectors_arguments()

Generate Job arguments for ExtractIvectorsFunction

Returns:

Arguments for processing

Return type:

list[ExtractIvectorsArguments]

property ie_path

Ivector extractor ie path

property plda_path

Path to trained PLDA model

property utterance_ivector_path

Path to scp file containing all ivectors