Recommended values for data augmentation

I saw the list of options available to perform data augmentation here:
https://mozilla-voice-stt.readthedocs.io/en/latest/TRAINING.html#augmentation
and was hoping that I can get some recommended values for the parameters of the augmentation stated in the docs.

Is it possible to specify what all augmentation values were used to train DeepSpeech v0.8.2? I checked the releases page, there wasn’t any mention of augmentation there.

As far as I know they are still experimenting with this feature. And my guess is that it wasn’t used for the current release. So you’ll probably have to find your own values.

1 Like