class montreal_forced_aligner.corpus.features.VadConfigMixin(energy_threshold=5.5, energy_mean_scale=0.5, **kwargs)[source]#

Bases: FeatureConfigMixin

Abstract mixin class for performing voice activity detection

  • use_energy (bool) – Flag for using the first coefficient of MFCCs

  • energy_threshold (float) – Energy threshold above which a frame will be counted as voiced

  • energy_mean_scale (float) – Proportion of the mean energy of the file that should be added to the energy_threshold

See also


For feature generation parameters

property vad_options#

Options for performing VAD