docker build failed with error: error loading package 'native_client'

DeepSpeech
#1

I am trying to build a docker image on my mac osx machine.
I am getting the following errors:

ERROR: Skipping ‘//native_client:libctc_decoder_with_kenlm.so’: error loading package ‘native_client’: Encountered error while reading extension file ‘cuda/build_defs.bzl’: no such package ‘@local_config_cuda//cuda’: Traceback (most recent call last):
…
Cuda Configuration Error: cuDNN version detected from /usr/lib/x86_64-linux-gnu/include/cudnn.h (7.1.4) does not match TF_CUDNN_VERSION (7.1.1)
WARNING: Target pattern parsing failed.
ERROR: error loading package ‘native_client’: Encountered error while reading extension file ‘cuda/build_defs.bzl’: no such package ‘@local_config_cuda//cuda’: Traceback (most recent call last):

I also see those two warnings before:
tensorflow 1.6.0 has requirement numpy>=1.13.3, but you’ll have numpy 1.11.0 which is incompatible.
tensorboard 1.6.0 has requirement numpy>=1.12.0, but you’ll have numpy 1.11.0 which is incompatible.

What should I do?
Thanks!

#2

Can you describe what you want to do ? This Docker file is for training with CUDA, so it’s very unlikely to give anything realistically working on macOS.

#3

Hi lissyx,

Thanks for your response.
We want to have a container running deepspeech docker, so we can run some voice-to-speech tasks and test the transcriptions quality.
I installed docker on my OSX machine and tried to build the deepspeech docker, in order to run it.

Any suggestions how to achieve it?

#4

If you are just looking to build some voice-to-text then you don’t need that Dockerfile. Just use the python or nodejs packages we have, or the C++ client from TaskCluster as documented in the README. Regarding Docker on OSX, I’m sorry but we cannot help on that.