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 369 forks source link

WER in speech recognition in inference mode #379

Closed inchpunch closed 5 years ago

inchpunch commented 5 years ago

Wondering where to find WER for the inference mode?

I tried the toy example and the small DS2 model

python run.py --config_file=example_configs/speech2text/ds2_toy_config.py --mode=infer --infer_output_file=ds2_toy_out.txt

python run.py --config_file=example_configs/speech2text/ds2_small_1gpu.py --mode=infer --infer_output_file=ds2_out.txt

But the screen output only says "*** Finished inference" without showing the WER. The output file does not contain WER either.

borisgin commented 5 years ago

WER is printed in the eval mode. WER for inference is not defined since we don't have ground truth