Building libdeepspeech.so for Android ARM64 in Windows

I’m trying to get DeepSpeech working in an Android app. I’m running Windows and am trying to follow the README instructions in the native_client directory of the Github repository.

So far I have managed to successfully download and install bazel, tensorflow r1.13 and create a symbolic link to the native_client folder from within the tensorflow directory.

However I’m running into errors building libdeepspeech.so with bazel:

C:\DeepSpeech-0.5.1\tensorflow>bazel build --workspace_status_command="bash native_client/bazel_workspace_status_cmd.sh" --config=monolithic --config=android --config=android_arm64 --define=runtime=tflite --action_env ANDROID_NDK_API_LEVEL=21 --cxxopt=-std=c++11 --copt=-D_GLIBCXX_USE_C99 //native_client:libdeepspeech.so
WARNING: The following rc files are no longer being read, please transfer their contents or import their path into one of the standard rc files:
c:\deepspeech-0.5.1\tensorflow/.bazelrc
WARNING: The following configs were expanded more than once: [monolithic, android]. For repeatable flags, repeats are counted twice and may lead to unexpected behavior.
WARNING: Option 'experimental_shortened_obj_file_path' is deprecated
ERROR: No toolchain found for cpu 'arm64-v8a'. Valid cpus from default_toolchain entries are: [
]. Valid toolchains are: [
  stub_armeabi-v7a: --cpu='armeabi-v7a' --compiler='compiler',
  msys_x64: --cpu='x64_windows' --compiler='msys-gcc',
  msys_x64_mingw: --cpu='x64_windows' --compiler='mingw-gcc',
  msvc_x64: --cpu='x64_windows' --compiler='msvc-cl',
]
INFO: Elapsed time: 13.913s
INFO: 0 processes.
FAILED: Build did NOT complete successfully (0 packages loaded)

I’d really appreciate any pointers on what might be causing this issue, and how I might be able to resolve the error.

I’m happy to provide further information about my system or the versions of tensorflow/bazel etc. installed if it would be helpful to diagnose.

@kw3rti Are you using our TensorFlow repo ? Also, you seem to build 0.5.1. Please follow the documentations and use r1.13 and not r1.14.

Yes and I’ve just checked, I’m actually using r1.13 and not r1.14 as I originally stated.

Have you properly ran ./configure ? The error really has nothing to do with DeepSpeech so far, it’s TensorFlow-level incomplete / broken setup.

Make sure you have properly followed upstream configuration steps for building for Android as well …