TensorFlow Android GRU Smart Prompter
This folder contains a voice-based word predictor application utilizing TensorFlow.
Description
This demo uses Google voice recognition service to translate voice to text and then uses a GRU (one variant of RNN) model to predict words from text history in real-time, displaying the top results in an overlay on the main activity.
To build/install/run
As a prerequisite, Bazel, the Android NDK, and the Android SDK, TensorFlow must all be installed on your system. And Google voice recognition service need to be installed on your smart phone.
- Get the recommended Bazel version listed at: https://www.tensorflow.org/versions/master/get_started/os_setup.html#source
- The Android NDK may be obtained from: http://developer.android.com/tools/sdk/ndk/index.html
- The Android SDK and build tools may be obtained from: https://developer.android.com/tools/revisions/build-tools.html
- The TensorFlow package can be found from: http://www.tensorflow.org
- Download any Google App with Google voice typing automatically install recognition service.
Then paste this directory under tensorflow/tensorflow/examples/.
The Android entries in <workspace_root>/WORKSPACE must be
uncommented with the paths filled in appropriately depending on where you
installed the NDK and SDK. Otherwise an error such as:
"The external label '//external:android/sdk' is not bound to anything" will
be reported.
In directory tensorflow/tensorflow/core/kernels, modify control_flow_op.cc: comment the three lines: "TF_CALL_ALL_TYPES(REGISTER_CPU_SWITCH)" "TF_CALL_ALL_TYPES(REGISTER_CPU_REF_SWITCH)" "TF_CALL_QUANTIZED_TYPES(REGISTER_CPU_SWITCH)" and add two lines above it: "REGISTER_KERNEL_BUILDER(Name("Switch").Device(DEVICE_CPU), SwitchOp)" "REGISTER_KERNEL_BUILDER(Name("RefSwitch".Device(DEVICE_CPU), SwitchOp))"
The model protobuf file and the vocab file are both already included in the assets directory
in the source tree. And the model definition file (GRU.py and run GRU.py) using tensorflow python API
are in the 'model directory.'
Then, after editing your WORKSPACE file, you must build the APK. Run this from
your workspace root:
$ bazel build //tensorflow/examples/android:word_predictorIf you get build errors about protocol buffers, run
git submodule update --init and build again.
If adb debugging is enabled on your Android 5.0 or later device, you may then use the following command from your workspace root to install the APK once built:
$ adb install -r -g bazel-bin/tensorflow/examples/android/word_predictor.apkSome older versions of adb might complain about the -g option (returning: "Error: Unknown option: -g"). In this case, if your device runs Android 6.0 or later, then make sure you update to the latest adb version before trying the install command again. If your device runs earlier versions of Android, however, you can issue the install command without the -g option.
Alternatively, a streamlined means of building, installing and running in one command is:
$ bazel mobile-install //tensorflow/examples/android:word_predictor --start_appIf voice record permission errors are encountered (possible on Android
Marshmallow or above), then the adb install command above should be used
instead, as it automatically grants the required voice permissions with -g.
The permission errors may not be obvious if the app halts immediately, so if
you installed with bazel and the app doesn't come up, then the easiest thing
to do is try installing with adb.
To use the Google voice recognition service, you need to connect your phone to the Internet running this app. Once the app is installed it will be named "Smart Prompter" and have the orange TensorFlow logo as its icon.