nnsvs.gen.gen_spsvs_static_features

nnsvs.gen.gen_spsvs_static_features(labels, acoustic_features, binary_dict, numeric_dict, stream_sizes, has_dynamic_features, pitch_idx=None, num_windows=3, frame_period=5, relative_f0=True, vibrato_scale=1.0, vuv_threshold=0.3, force_fix_vuv=True)[source]

Generate static features from predicted acoustic features

Parameters:

  • labels (HTSLabelFile) – HTS labels

  • acoustic_features (ndarray) – predicted acoustic features

  • binary_dict (dict) – binary feature dictionary

  • numeric_dict (dict) – numeric feature dictionary

  • stream_sizes (list) – stream sizes

  • has_dynamic_features (list) – whether each stream has dynamic features

  • pitch_idx (int) – index of pitch features

  • num_windows (int) – number of windows

  • frame_period (float) – frame period

  • relative_f0 (bool) – whether to use relative f0

  • vibrato_scale (float) – vibrato scale

  • vuv_threshold (float) – vuv threshold

  • force_fix_vuv (bool) – whether to use post-processing to fix VUV.

Returns:

tuple of mgc, lf0, vuv and bap.

Return type:

tuple