There are some languages where an entire utterance might not have pitch/periodicity. In these cases we should not throw an error as we used to do when interpolating nan pitch values. However, more often than not, pitch-less audio actually represents bad data, so we should warn the user about this while preprocessing somehow.
Related to #284 and #308
There are some languages where an entire utterance might not have pitch/periodicity. In these cases we should not throw an error as we used to do when interpolating nan pitch values. However, more often than not, pitch-less audio actually represents bad data, so we should warn the user about this while preprocessing somehow.
Related to #284 and #308