modify RecorderSpeechRecognizer, set default speex encoding to C++ lib

Author: olami-developers

lib/build.gradle

@@ -8,10 +8,11 @@ android {
         minSdkVersion 17
         targetSdkVersion 25
         versionCode 1
-        versionName "2.5.0"
+        versionName "2.5.1"
 
         archivesBaseName = "olami-android-client"
         version = android.defaultConfig.versionName
+
         externalNativeBuild {
             cmake {
                 cppFlags "-std=c++11 -frtti -fexceptions"
@@ -36,12 +37,12 @@ android {
         }
     }
 
+    sourceSets { main { jni.srcDirs = ['src/main/jni', 'src/main/jni/'] } }
+
     lintOptions {
         abortOnError false
     }
 
-    sourceSets { main { jni.srcDirs = ['src/main/jni', 'src/main/jni/'] } }
-
     libraryVariants.all { variant ->
         variant.outputs.each { output ->
             def outputFile = output.outputFile

lib/src/main/java/ai/olami/android/KeepRecordingSpeechRecognizer.java

@@ -82,9 +82,10 @@ public class KeepRecordingSpeechRecognizer extends SpeechRecognizerBase {
 
     private VoiceVolume mVoiceVolume = new VoiceVolume();
 
-    private RecognizeState mRecognizeState = null;
     private Codec mSpeexEncoder = null;
 
+    private RecognizeState mRecognizeState = null;
+
     /**
      * Recognize process state
      */

lib/src/main/java/ai/olami/android/RecorderSpeechRecognizer.java

@@ -32,6 +32,7 @@
 import java.util.concurrent.BlockingQueue;
 import java.util.concurrent.LinkedBlockingQueue;
 
+import ai.olami.android.jni.Codec;
 import ai.olami.cloudService.APIConfiguration;
 import ai.olami.cloudService.APIResponse;
 import ai.olami.cloudService.CookieSet;
@@ -75,8 +76,10 @@ public class RecorderSpeechRecognizer extends SpeechRecognizerBase{
 
     private VoiceVolume mVoiceVolume = new VoiceVolume();
 
-    private RecordState mRecordState;
-    private RecognizeState mRecognizeState;
+    private Codec mSpeexEncoder = null;
+
+    private RecordState mRecordState = null;
+    private RecognizeState mRecognizeState = null;
 
     /**
      * Recording state
@@ -482,7 +485,19 @@ private void doSending() throws Exception {
                 byte[] audioData = (byte[]) mRecordDataQueue.take();
                 mIsFinal = (isRecodingStopped() && (mRecordDataQueue.isEmpty()));
                 length += ((audioData.length / getFrameSize()) * FRAME_LENGTH_MILLISECONDS);
-                mRecognizer.appendAudioFramesData(audioData);
+                if (getAudioCompressLibraryType() == AUDIO_COMPRESS_LIBRARY_TYPE_CPP) {
+                    if (mSpeexEncoder == null) {
+                        mSpeexEncoder = new Codec();
+                        mSpeexEncoder.open(1, 10);
+                    }
+                    byte[] encBuffer = new byte[audioData.length];
+                    int encSize = mSpeexEncoder.encodeByte(audioData, 0, audioData.length, encBuffer);
+                    mRecognizer.setAudioType(SpeechRecognizer.AUDIO_TYPE_PCM_SPEEX);
+                    mRecognizer.appendSpeexAudioFramesData(encBuffer, encSize);
+                } else {
+                    mRecognizer.setAudioType(SpeechRecognizer.AUDIO_TYPE_PCM_RAW);
+                    mRecognizer.appendAudioFramesData(audioData);
+                }
                 if ((length >= getUploadAudioLengthMilliseconds()) || mIsFinal) {
                     APIResponse response = mRecognizer.flushToUploadAudio(mCookie, mIsFinal);
                     if (response.ok()) {
@@ -514,6 +529,11 @@ private void doSending() throws Exception {
             }
         }
 
+        if (mSpeexEncoder != null) {
+            mSpeexEncoder.close();
+            mSpeexEncoder = null;
+        }
+
         synchronized (mRecordDataQueue) {
             mRecordDataQueue.clear();
             mRecordDataQueue = null;