NameError: name 'psf' is not defined

NVIDIA / OpenSeq2Seq

Toolkit for efficient experimentation with Speech Recognition, Text2Speech and NLP

https://nvidia.github.io/OpenSeq2Seq

Apache License 2.0

1.54k stars 371 forks source link

NameError: name 'psf' is not defined #432

Closed vinnitu closed 5 years ago

vinnitu commented 5 years ago

how to define psf?

now I add backend librosa to data_layer_params

vsl9 commented 5 years ago

Did you install python_speech_features package? We started development with python_speech_features and then switched to librosa. By default, backend is psf (python_speech_features) for backward compatibility, so there is no need to define psf backend. Just make sure that you installed python_speech_features. Though, we recommend to use newer librosa backend which has numerous important features (like windowing, more accurate mel scale aggregation, etc.).