pbaljeka / world_merlin Goto Github PK

I have build a model for indic Hindi language. It resynthesizing part is working well on test data.
But I am still curious, how the trained model resided in "gen" folder is going to be used for generating voice on new text.

thank you!!

While running setup_world_merlin, error message appear because wagon_rf_f0 doesn't exist in clustergen.

I am trying to run this recipe for indic marathi voice.

I am getting this error:

2018-02-19 18:26:34,860 INFO main : normalising acoustic (output) features using method MVN
Traceback (most recent call last):
File "ss_dnn/merlin_scripts/src/run_dnn.py", line 983, in
main_function(cfg)
File "ss_dnn/merlin_scripts/src/run_dnn.py", line 691, in main_function
nn_cmp_norm_file_list[0:int(cfg.train_file_number)+int(cfg.valid_file_number)])
File "/mnt/disks/disk1/gits/cmu_indic_mar_msv/ss_dnn/merlin_scripts/src/frontend/mean_variance_norm.py", line 69, in feature_normalisation
if self.mean_vector == None:
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

....
please let me know what more details I should provide
Thanks