Support AAudio based input recording to enable full duplex AAudio sup…

…port

* Initial implementation of AAudio based input recording
* Verified duplex operation on Pixel 3
* Made unit test mode working with openSL mock, as before
* Updated example activity
* Fix handlers
* Buffer size and temporary buffers now update during runtime buffer size updates. Use preferred floating point format for AAUdio, use memory cloning for faster writes.
* README and inline doc copy updates
* and again
* Fall back to OpenSL default

.gitignore

@@ -10,4 +10,5 @@ src/main/cpp/jni/java_interface_wrap.cpp
 libs/*
 obj
 build
+debug
 local.properties

README.md

@@ -2,9 +2,7 @@ MWEngine is..
 =============
 
 ...an audio engine for Android, using either OpenSL (compatible with Android 4.1 and up) or AAudio
-(Android 8.0 and up) as the drivers for low latency audio performance. The engine has been written for both
-[MikroWave](https://play.google.com/store/apps/details?id=nl.igorski.mikrowave.free&hl=en) and
-[Kosm](https://play.google.com/store/apps/details?id=nl.igorski.kosm&hl=en) to provide fast live audio synthesis. MWEngine is also used by [TIZE - Beat Maker, Music Maker](https://play.google.com/store/apps/details?id=com.tizemusic.tize).
+(Android 8.0 and up) as the drivers for low latency audio performance.
 
 MWEngine provides an architecture that allows you to work with audio within a _musical context_. It is easy to
 build upon the base classes and create your own noise generating mayhem. A few keywords describing the
@@ -16,21 +14,29 @@ out-of-the-box possibilities are:
  * effect chains operating on individual input/output channels
  * sample playback with real time pitch shifting
  * bouncing output to WAV files, either live (during a performance) or "offline"
- 
+
 Also note that MWEngine's underlying audio drivers are _the same as Google Oboe uses_, MWEngine and
 Oboe are merely abstraction layers to solve the same problem, only in different ways. Additionally, MWEngine provides a complete audio processing environment.
 
+#### Who uses this ?
+
+The engine has been written for both [MikroWave](https://play.google.com/store/apps/details?id=nl.igorski.mikrowave.free&hl=en) and
+[Kosm](https://play.google.com/store/apps/details?id=nl.igorski.kosm&hl=en) to provide fast live audio synthesis.
+
+While developments on those apps are scarce, the engine itself has been continuously improved and is now also
+used by third party app developers, such as [TIZE - Beat Maker, Music Maker](https://play.google.com/store/apps/details?id=com.tizemusic.tize).
+
 ### The [Issue Tracker](https://github.com/igorski/MWEngine/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-desc) is your point of contact
 
 Bug reports, feature requests, questions and discussions are welcome on the GitHub Issue Tracker, please do not send e-mails through the development website. However, please search before posting to avoid duplicates, and limit to one issue per post.
 
 Please vote on feature requests by using the Thumbs Up/Down reaction on the first post.
 
-### C++ ??? What about Java ?
+### C++ ??? What about Java / Kotlin ?
 
 Though the library is written in C++ (and can be used solely within this context), the library can be built using JNI
 (Java Native Interface) which makes its API expose itself to Java, while still executing in a native layer outside of
-the Dalvik/ART VM. In other words : high performance of the engine is ensured by the native layer operations, while
+the JVM. In other words : high performance of the engine is ensured by the native layer operations, while
 ease of development is ensured by delegating application logic / UI to the realm of the Android Java SDK.
 
 Whether you intend to use MWEngine for its sample based playback or to leverage its built-in synthesizer and
@@ -139,19 +145,16 @@ sequence going using the library.
 To install the demo: first build the library as described above, and then run the build script to deploy the .APK onto an
 attached device/emulator (note that older emulated devices can only operate at a sample rate of 8 kHz!).
 
-### Note on AAudio
+### Note on OpenSL / AAudio drivers
+
+Currently it is not possible to switch between audio drivers on the fly, rather you must precompile
+the library for use with a specific driver. By default, the library will compile for OpenSL for a
+wider range of supported devices. If you want to use AAudio instead (and thus are targeting solely
+devices running Android 8 and up) :
 
-The AAudio implementation has been built using (in Google's words): _"a Preview release of the AAudio library. The API
-might change in backward-incompatible ways in future releases. It is not recommended for production use."_ so use it
-at your own peril. To use AAudio instead of OpenSL:
-
  * change the desired driver in _global.h_ from type 0 (OpenSL) to 1 (AAudio)
- * update the _Android.mk_ file to include all required adapters and libraries (simply set _BUILD_AAUDIO_ to 'true')
- * update target in _project.properties_ to _android-26_
-
-Once AAudio is a stable library, MWEngine will allow on-the-fly switching between OpenSL and AAudio drivers.
 
-(!) MWEngine does not support recording from the device inputs using AAudio just yet, (https://github.com/igorski/MWEngine/issues/70) references this feature.
+Should you require support for both driver variants, please file a feature request in the repository's issue tracker.
 
 ### Contributors
 

build.gradle

@@ -15,7 +15,7 @@ android {
 
     defaultConfig {
         applicationId "nl.igorski.example"
-        minSdkVersion 16
+        minSdkVersion 26 // can go down to 16 when using OpenSL as the audio driver
         targetSdkVersion 27
         versionCode 1
         versionName "1.0.0"

src/main/cpp/Android.mk

@@ -1,6 +1,3 @@
-# Experimental AAudio support, set to true when building for AAudio (requires NDK target 26)
-BUILD_AAUDIO = false
-
 LOCAL_PATH      := $(call my-dir)
 LOCAL_SRC_FILES := \
 
@@ -32,6 +29,7 @@ global.cpp \
 jni/javabridge.cpp \
 drivers/adapter.cpp \
 drivers/opensl_io.c \
+drivers/aaudio_io.cpp \
 utilities/utils.cpp \
 audioengine.cpp \
 audiobuffer.cpp \
@@ -94,14 +92,7 @@ modules/envelopefollower.cpp \
 modules/lfo.cpp \
 modules/routeableoscillator.cpp \
 
-ifeq ($(BUILD_AAUDIO),true)
-    LOCAL_SRC_FILES   += \
-    drivers/aaudio_io.cpp \
-
-    LOCAL_LDLIBS := -laaudio
-endif
-
-LOCAL_LDLIBS += -lOpenSLES -landroid -latomic -llog
+LOCAL_LDLIBS += -lOpenSLES -laaudio -landroid -latomic -llog
 
 include $(BUILD_SHARED_LIBRARY)
 

src/main/cpp/Application.mk

@@ -3,6 +3,7 @@ APP_STL        := c++_static
 APP_CPPFLAGS   += -std=c++11 -Werror -fexceptions -frtti
 #APP_CPPFLAGS   += -Wall
 APP_ABI        := x86 x86_64 armeabi-v7a arm64-v8a
+APP_PLATFORM   = android-26
 
 ifeq ($(TARGET_ARCH_ABI), x86)
 LOCAL_CFLAGS += -m32

src/main/cpp/audioengine.cpp

@@ -151,8 +151,9 @@ namespace MWEngine {
         // generate the input buffer used for recording from the device's input
         // as well as the temporary buffer used to merge the input into
 
-        recbufferIn  = new float[ AudioEngineProps::BUFFER_SIZE ]();
+        recbufferIn  = new float[ AudioEngineProps::BUFFER_SIZE * AudioEngineProps::INPUT_CHANNELS ]();
         inputChannel->createOutputBuffer();
+
 #endif
         // accumulates all channels ("master strip")
 
@@ -260,17 +261,19 @@ namespace MWEngine {
         // record audio from Android device ?
         if (( recordDeviceInput || recordInputToDisk ) && AudioEngineProps::INPUT_CHANNELS > 0 )
         {
-            int recSamps                  = DriverAdapter::getInput( recbufferIn );
+            int recordedSamples           = DriverAdapter::getInput( recbufferIn, amountOfSamples );
             SAMPLE_TYPE* recBufferChannel = inputChannel->getOutputBuffer()->getBufferForChannel( 0 );
 
-            for ( int j = 0; j < recSamps; ++j )
+            for ( int j = 0; j < recordedSamples; ++j ) {
                 recBufferChannel[ j ] = recbufferIn[ j ];//static_cast<float>( recbufferIn[ j ] );
+            }
 
             // apply processing chain onto the input
 
             std::vector<BaseProcessor*> processors = inputChannel->processingChain->getActiveProcessors();
-            for ( int k = 0; k < processors.size(); ++k )
+            for ( int k = 0; k < processors.size(); ++k ) {
                 processors[ k ]->process( inputChannel->getOutputBuffer(), AudioEngineProps::INPUT_CHANNELS == 1 );
+            }
 
             // merge recording into current input buffer for instant monitoring
 
@@ -373,8 +376,9 @@ namespace MWEngine {
             }
 
             // write cache if it didn't happen yet ;) (bus processors are (currently) non-cacheable)
-            if ( mustCache )
+            if ( mustCache ) {
                 mustCache = !writeChannelCache( channel, channelBuffer, cacheReadPos );
+            }
 
             // write the channel buffer into the combined output buffer, apply channel volume
             // note live events are always audible as their volume is relative to the instrument