Sample Android App for JNI library

.gitignore

@@ -8,9 +8,42 @@ cache/
 external/
 test/dataset/
 
+.vscode/
+
 # Never exclude src files
 !WhisperKit/**
 !cli/**
 
 # MacOS specific
 .DS_Store
+
+### Android ###
+# IntelliJ IDEA project files
+.idea/
+*.iml
+*.ipr
+*.iws
+out/
+
+# Model folder (too large)
+androidApp/app/src/main/assets/openai_whisper-tiny/
+
+# Gradle files and caches
+.gradle/
+local.properties
+
+# Native build outputs (e.g., C++/JNI)
+.cxx/
+
+# Heap dump files
+*.hprof
+
+# Log files
+*.log
+
+# test audio files
+androidApp/app/src/main/assets/*.wav
+androidApp/app/src/main/assets/*.m4a
+
+# .so folder in the app (repeated files)
+androidApp/app/src/main/jniLibs

README.md

@@ -21,6 +21,7 @@ but we are continuing to invest in Android and now welcome contributions from th
 - [Installation](#installation)
 - [Getting Started](#getting-started)
 - [CLI Run and Test](#cli-run-and-test)
+- [Running Android app](#running-android-app)
 - [Contributing \& Roadmap](#contributing--roadmap)
 - [License](#license)
 - [Citation](#citation)
@@ -74,12 +75,13 @@ make env
 
 2. Inside the Docker environment, build the `whisperkit-cli` CLI using (for Android and Linux):
 ```
-make build [linux | qnn | gpu]
+make build [linux | qnn | gpu | jni]
 ```
 
 The QNN option builds WhisperKit with Qualcomm AI NPU support and the QNN TFLite delegate.
 The 'gpu' option is the generic GPU backend for all Android devices from TFLite GPU delegate.
 Linux builds are currently CPU-only.
+The 'jni' option builds the .so file with JNI library to use on android (using QNN support).
 
 3. Back on the host machine (outside Docker shell), push dependencies to the Android device:
 ```
@@ -110,7 +112,16 @@ For Linux:
 make build linux
 ```
 
-2. Manually run `whisperkit-cli`:
+2. Run on Android with `run_on_android.sh` script:
+
+Log in via adb shell:
+```
+adb shell
+cd /sdcard/argmax/tflite
+sh run_on_android.sh
+```
+
+3. Manually run `whisperkit-cli`:
 
 Usage: 
 
@@ -129,7 +140,7 @@ export LD_LIBRARY_PATH=/data/local/tmp/lib
 whisperkit-cli transcribe --model-path  /path/to/openai_whisper-base --audio-path /path/to/inputs/jfk_441khz.m4a
 ```
 
-3. Sample execution output:
+4. Sample execution output:
 ```
 root@cf40510e9b93:/src/AXIE# ./build/linux/whisperkit-cli transcribe --model-path /src/AXIE/models/openai_whisper-small --audio-path /src/AXIE/test/jfk_441khz.m4a 
 SoC: 	generic CPU (x86, arm64, etc) 
@@ -153,6 +164,22 @@ Transcription:   And so, my fellow Americans, ask not what your country can do f
 ```
 </details>
 
+# Running Android App
+
+<details>
+  <summary> (Click to expand) </summary>
+
+1. Move model to assets folder  
+Download the models if you haven't done so as specified in the Installation section. Move specifically the "whisper_tiny" folder into the assets folder of the app. 
+
+2. Run the app with Android Studio
+
+3. Things to consider.  
+QNN will only work if the SoC of your device is among the supported ones listed in the C++ code. The model may take a couple minutes to load when using QNN delegate.
+After recording with the microphone, the input is saved into the MicInput.wav file, you can select it to transcribe your audio.
+
+</details>
+
 ## Contributing
 
 WhisperKit Android is currently in the v0.1 Beta stage.  We are actively developing the project and welcome contributions from the community.

WhisperKit/src/TranscribeTask.cpp

@@ -30,7 +30,7 @@ extern "C" {
 #define TRANSCRIBE_TASK_DEFAULT_LIB_DIR     "/data/local/tmp/lib"
 #define TRANSCRIBE_TASK_DEFAULT_CACHE_DIR   "/data/user/0/com.whispertflite/cache"
 #elif (QNN_DELEGATE || GPU_DELEGATE) 
-// for Android QNN or GPU delegatea
+// for Android QNN or GPU delegate
 #define TRANSCRIBE_TASK_TFLITE_ROOT_PATH    "/sdcard/argmax/tflite"
 #define TRANSCRIBE_TASK_DEFAULT_LIB_DIR     "/data/local/tmp/lib"
 #define TRANSCRIBE_TASK_DEFAULT_CACHE_DIR   "/data/local/tmp/cache"
@@ -199,7 +199,7 @@ std::unique_ptr<std::string> Runtime::cmdexec(const char* cmd) {
 
 bool Runtime::check_qcom_soc() {
     vector<string> supported_socs{
-        "SM8650", "SM8550", "SM8450","SM8350"
+        "SM8650", "SM8550", "SM8450","SM8350", "SM7450"
     };
 
     auto soc = *cmdexec("getprop ro.soc.model");

androidApp/app/build.gradle

@@ -0,0 +1,72 @@
+plugins {
+    id 'com.android.application'
+    id 'org.jetbrains.kotlin.android'
+}
+
+android {
+    signingConfigs {
+        release {
+        }
+    }
+    namespace 'com.whispertflite'
+    compileSdk 34
+
+    defaultConfig {
+        applicationId "com.whispertflite"
+        minSdk 26
+        targetSdk 34
+        versionCode 1
+        versionName "1.0"
+
+        testInstrumentationRunner "androidx.test.runner.AndroidJUnitRunner"
+
+        // Specify the ABIs to exclude
+        ndk {
+            abiFilters 'armeabi-v7a', 'arm64-v8a'
+        }
+    }
+
+    buildTypes {
+        release {
+            minifyEnabled false
+            proguardFiles getDefaultProguardFile('proguard-android-optimize.txt'), 'proguard-rules.pro'
+            signingConfig signingConfigs.debug
+        }
+    }
+
+    buildFeatures {
+        compose true
+    }
+    composeOptions {
+        kotlinCompilerExtensionVersion = '1.5.14'
+    }
+//    externalNativeBuild {
+//        cmake {
+//            path "src/main/cpp/WhisperKitAndroid/CMakeLists.txt"
+//        }
+//    }
+    compileOptions {
+        sourceCompatibility = JavaVersion.VERSION_11
+        targetCompatibility = JavaVersion.VERSION_11
+    }
+    kotlinOptions {
+        jvmTarget = '11'
+    }
+}
+
+dependencies {
+    implementation 'androidx.appcompat:appcompat:1.7.0'
+    implementation 'com.google.android.material:material:1.12.0'
+    implementation 'androidx.constraintlayout:constraintlayout:2.2.0'
+
+    // Main TensorFlow Lite library
+    implementation 'org.tensorflow:tensorflow-lite:2.14.0'
+    implementation 'org.tensorflow:tensorflow-lite-support:0.4.4'
+    implementation 'com.google.android.gms:play-services-tflite-gpu:16.4.0'
+    implementation 'com.google.android.gms:play-services-tflite-java:16.4.0'
+    implementation 'androidx.compose.ui:ui-android:1.7.6'
+    implementation 'androidx.compose.foundation:foundation-android:1.7.6'
+    implementation 'androidx.compose.material3:material3-android:1.3.1'
+    implementation 'androidx.activity:activity-compose:1.9.0'
+
+}

androidApp/app/proguard-rules.pro

@@ -0,0 +1,21 @@
+# Add project specific ProGuard rules here.
+# You can control the set of applied configuration files using the
+# proguardFiles setting in build.gradle.
+#
+# For more details, see
+#   http://developer.android.com/guide/developing/tools/proguard.html
+
+# If your project uses WebView with JS, uncomment the following
+# and specify the fully qualified class name to the JavaScript interface
+# class:
+#-keepclassmembers class fqcn.of.javascript.interface.for.webview {
+#   public *;
+#}
+
+# Uncomment this to preserve the line number information for
+# debugging stack traces.
+#-keepattributes SourceFile,LineNumberTable
+
+# If you keep the line number information, uncomment this to
+# hide the original source file name.
+#-renamesourcefileattribute SourceFile

androidApp/app/src/main/AndroidManifest.xml

@@ -0,0 +1,35 @@
+<?xml version="1.0" encoding="utf-8"?>
+<manifest xmlns:android="http://schemas.android.com/apk/res/android"
+    xmlns:tools="http://schemas.android.com/tools"
+    package="com.whispertflite">
+
+    <uses-permission android:name="android.permission.RECORD_AUDIO" />
+    <application
+        android:allowBackup="true"
+        android:icon="@mipmap/ic_launcher"
+        android:label="WhisperASR"
+        android:roundIcon="@mipmap/ic_launcher_round"
+        android:supportsRtl="true"
+        android:extractNativeLibs="true"
+        tools:targetApi="31">
+        <uses-native-library
+            android:name="libcdsprpc.so"
+            android:required="false"/>
+        <activity android:name="com.whispertflite.MainActivity" android:exported="true">
+            <intent-filter>
+                <action android:name="android.intent.action.MAIN" />
+                <category android:name="android.intent.category.LAUNCHER" />
+            </intent-filter>
+
+        </activity>
+        <activity
+            android:name="org.libsdl.app.SDLActivity"
+            android:label="SDL"
+            android:launchMode="singleTask"
+            android:exported="false"
+            >
+
+        </activity>
+    </application>
+
+</manifest>