README.md

An Android app for speech recognition based on TensorFlow library.

Step 1. Software requirement

Step 1.1 Install Android Studio 2.3 +

Android Studio (Build tools API >= 23 is required to build the TF Android demo (though it will run on API >= 21 devices).)

https://developer.android.com/studio/

'''(Skip to Step2.)''' Step 1.2 Download TensorFlow inference library

https://ci.tensorflow.org/view/Nightly/job/nightly-android/

download two files:

*libandroid_tensorflow_inference_java.jar

refer to: https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/android

*libtensorflow_inference.so

includes : tensorflow-inference-debug.aar and tensorflow-inference-release.aar;

refer to: https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/android/cmake

(optinal) Step 2. Training Model and Frozen graph (Work done on Project AudioNet )

Using audioNet to foster a *.h5 file and frozen as *.pb file

(optinal) Step 3. Put model file in android project (already)

put ‘asrModel.pb’ in assets folder

(optinal) Step 4. Run Tensorflow inference interface (already intergated into project)

Step 5. Run Android Studio, open this project: androidAudioRecg

---build apk file, Install apk file in your mobile phone.

done!

Reference

see Tensorflow : An open-source software library for Machine Intelligence.

https://www.tensorflow.org