search

NVIDIA / OpenSeq2Seq

Toolkit for efficient experimentation with Speech Recognition, Text2Speech and NLP
https://nvidia.github.io/OpenSeq2Seq
Apache License 2.0
1.54k stars 372 forks source link

Unreadable Output #538

Open KareemAtef123 opened 4 years ago

KareemAtef123 commented 4 years ago

Hello guys, I am using deep speech 2 pre-trained model and then my infer output (light.txt) of the samples.csv (inputs) is an encoded text that is not readable. Please tell me how may I solve this problem? unreadable output

KareemAtef123 commented 4 years ago

For those who had the same problem, the solution was to set the 'infer_logits_to_pickle': to False in the decoder parameters.