Evaluating speech model does not generate any output! #86
Description
I have generated a speech model using kur, speech.yml. Now I am trying to evaluate my model by adding the following line in speech.yml.
But the output pickle file contains 'None' as truth and 'constant value' as prediction when there are multiple audio in the test data. Moreover, when I provide a single audio file in test data, it shows 'Keyerror' while I am reading this pickle file using the following code-
import pickle
import numpy
with open('./output/output.pkl', 'rb') as fh:
data = pickle.loads(fh.read())
#print(list(data.values()))
print(data)What could be the problem? any help would be appreciated..