Hi
I getting an error when running the native client (deepspeech) built from master. I don’t get this error with the one from v0.5 alpha 4. This was tested on debian running “testing” up to date.
I’ve compiled the libdeepspeech.so following the instructions.The steps were
- Checkout
https://github.com/mozilla/tensorflow - Install bazel from bazel-0.19.2-installer-linux-x86_64.sh
- bazel build -c opt --copt=-O3 //tensorflow:libtensorflow_cc.so
Target //tensorflow:libtensorflow_cc.so up-to-date:
bazel-bin/tensorflow/libtensorflow_cc.so
INFO: Elapsed time: 2377.256s, Critical Path: 205.47s, Remote (0.00% of the time): [queue: 0.00%, setup: 0.00%, process: 0.00%]
INFO: 6028 processes: 6028 local.
INFO: Build completed successfully, 6205 total actions
- bazel build -c opt --copt=-O3 //native_client:libdeepspeech.so
(had to fix an issue with “Darwin” in the BUILD file)
Target //native_client:libdeepspeech.so up-to-date:
bazel-bin/native_client/libdeepspeech.so
INFO: Elapsed time: 135.629s, Critical Path: 91.75s, Remote (0.00% of the time): [queue: 0.00%, setup: 0.00%, process: 0.00%]
INFO: 463 processes: 463 local.
INFO: Build completed successfully, 475 total actions
- Went to “native_client” directory and built with “make”, with success.
c++ -std=c++11 -o deepspeech `pkg-config --cflags sox` client.cc -Wl,--no-as-needed -Wl,-rpath,\$ORIGIN -L/home/miguel/Development/IPL/investigacao/tensorflow/bazel-bin/native_client -ldeepspeech `pkg-config --libs sox`
Then running the executable I get an error:
make ARGS="--model /home/miguel/Development/IPL/investigacao/dsmodel/models/output_graph.pb --trie /home/miguel/Development/IPL/investigacao/dsmodel/models/trie --alphabet /home/miguel/Development/IPL/investigacao/dsmodel/models/alphabet.txt --audio meco-44.1kHz.wav" run |& head
LD_LIBRARY_PATH=/home/miguel/Development/IPL/investigacao/tensorflow/bazel-bin/native_client: ./deepspeech --model /home/miguel/Development/IPL/investigacao/dsmodel/models/output_graph.pb --trie /home/miguel/Development/IPL/investigacao/dsmodel/models/trie --alphabet /home/miguel/Development/IPL/investigacao/dsmodel/models/alphabet.txt --audio meco-44.1kHz.wav
TensorFlow: v1.12.0-10463-g892ffbc922
DeepSpeech: v0.5.0-alpha.4-35-g25a254f
Warning: reading entire model file into memory. Transform model file into an mmapped graph to reduce heap usage.
2019-04-04 19:37:15.270567: I external/org_tensorflow/tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
2019-04-04 19:37:15.418865: E external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1355] OpKernel ('op: "RandomShuffle" device_type: "CPU" constraint { name: "T" allowed_values { list { type: DT_VARIANT } } }') for unknown op: RandomShuffle
2019-04-04 19:37:15.418883: E external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1355] OpKernel ('op: "RandomShuffle" device_type: "CPU" constraint { name: "T" allowed_values { list { type: DT_RESOURCE } } }') for unknown op: RandomShuffle
2019-04-04 19:37:15.418888: E external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1355] OpKernel ('op: "RandomShuffle" device_type: "CPU" constraint { name: "T" allowed_values { list { type: DT_STRING } } }') for unknown op: RandomShuffle
2019-04-04 19:37:15.418893: E external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1355] OpKernel ('op: "RandomShuffle" device_type: "CPU" constraint { name: "T" allowed_values { list { type: DT_BOOL } } }') for unknown op: RandomShuffle
2019-04-04 19:37:15.418897: E external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1355] OpKernel ('op: "RandomShuffle" device_type: "CPU" constraint { name: "T" allowed_values { list { type: DT_COMPLEX128 } } }') for unknown op: RandomShuffle
... many more lines
2019-04-04 19:37:40.955635: E external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1355] OpKernel ('op: "Enter" device_type: "GPU" constraint { name: "T" allowed_values { list { type: DT_BFLOAT16 } } }') for unknown op: Enter
2019-04-04 19:37:40.955639: E external/org_tensorflow/tensorflow/core/framework/op_kernel.cc:1355] OpKernel ('op: "Enter" device_type: "GPU" constraint { name: "T" allowed_values { list { type: DT_HALF } } }') for unknown op: Enter
Not found: Op type not registered 'VariableV2' in binary running on legion. Make sure the Op and Kernel are registered in the binary running in this process. Note that if you are loading a saved graph which used ops from tf.contrib, accessing (e.g.) `tf.contrib.resampler` should be done before importing the graph, as contrib ops are lazily registered when the module is first accessed.
Could not create model.
make: *** [Makefile:29: run] Error 1
Any help is appreciated.
Best regards,
Miguel Negrão