What happens if we don't specify a probabilty value for augmentation in training?

Archives DeepSpeech
#1

In the documentation, the format for specifying the pitch augmentation is as follows:

Pitch augmentation --augment pitch[p=<float>,pitch=<float-range>]

In the release 0.9.3, from the release notes, the following pitch parameter is used:

augment pitch[pitch=1~0.1]

As there is no probability value specified, does that mean p=1? because I couldn’t find the default value anywhere

#2

Hi, I don’t know the answer, but you might try on Mozilla’s Matrix channel for the DeepSpeech team.