nnsvs.gen

Generation

Generation from trained models

predict_timelag	Predict time-lag from HTS labels
predict_duration	Predict phoneme durations from HTS labels
predict_timing	Predict timinigs from HTS labels
predict_acoustic	Predict acoustic features from HTS labels
predict_waveform	Predict waveform from multi-stream acoustic features

Post-processisng

postprocess_duration	Post-process durations based on predicted time-lag
postprocess_acoustic	Post-process acoustic features
postprocess_waveform	Perform post-processing for synthesized waveform

Generation helpers

correct_vuv_by_phone	Correct V/UV by phone-related flags in a hed file
gen_spsvs_static_features	Generate static features from predicted acoustic features
gen_world_params	Generate WORLD parameters from mgc, lf0, vuv and bap.

Utils

get_windows

Get windows for MLPG.