DeepSpeech training - Data too short when trying to read string error

[saishruthi.tn]

Hello,

I am trying to train deepspeech model using common voice dataset. Below are the points where I got stuck. Any help on this will be useful.

1. Error
   tensorflow.python.framework.errors_impl.InvalidArgumentError: Data too short when trying to read string

2. How to restrict the number of steps taken for every epoch?
   Could not find which argument controls this.

Thanks in advance.

[areyliu6 (Arey)]

I’m not pretty sure about the first question, for the second question, the number of steps is equal to the amount of data divide by batch size so that you just need to increase the batch size (but the batch size is an important value for getting the gradient, may cause some problem if the batch size too big).

[saishruthi.tn]

Thanks @areyliu6 for the response

[April_yq (Longisland)]

@saishruthi.tn I face the same error as the first one.Did you work out the problem?