Standard Method for Processing Long Audio Files with 0.3.0/0.4.0 Python Package?

DeepSpeech
#21

I’ll look into that. Here is the link to the models I’m downloading in my Dockerfile: https://github.com/mozilla/DeepSpeech/releases/download/v0.3.0/deepspeech-0.3.0-models.tar.gz

#22

Except that the version you are running with is really important. …

#23

Alright, so we’re talking about the output after running ./deepspeech-0.3.0-models/libdeepspeech.so ?

#24

Sorry, I haven’t been able to find that file. It seems to be related to the python package in some way (found mentions of Native Client in the git repo)?

#25

We package that inside the Python wheel, but you cannot directly execute it.

#26

@zaptrem Any call to DS_CreateModel() will print versions informations on stderr: https://github.com/mozilla/DeepSpeech/blob/master/native_client/deepspeech.cc#L356. So you should be able to get it, even with the snippet you posted earlier.

#27

Bit of a delay, but here you go:

TensorFlow: v1.11.0-9-g97d851f
DeepSpeech: v0.3.0-0-gef6b5bd

#28

Thanks! So if you are confident this is matching the version that runs and produces the bad output, then it’s not the issue I had in mind of mixing 0.3.0 model with binaries build after 0.3.0.