examples directory contains some programs that demonstrate the functionality
of the engine, its interfaces and the suggested practices for a correct and efficient
use. There are currently examples for Linux, Windows and Android. Other platforms
may be supported in the future, but there is no immediate plan for that.
Linux and Windows examples¶
Example programs for Linux and Windows can be found in the
In order for the examples to build it is necessary to properly set the include and lib directories in the User Config section of the main CMake script (CMakeLists.txt) in the root directory.
All of the examples use an implementation of the
DataStore interface based on
key-value databases. These implementations can be found in the
The following drivers are provided:
TCDataStore- This driver provides access to Tokyo Cabinet, a memory-mapped, key-value store management library. It is an “embedded database”, so that’s the right choice for deploying ACR applications on embedded or mobile devices for on-device recognition (that is without the need to access a database over a network).
CBDataStore- This driver provides access to Couchbase, a distributed key-value/document database for high-performance applications. This is the typical database based on a client-server architecture and is suitable for large scale applications (e.g. web services). The examples have been tested on the Community Edition.
The examples only need one of those databases (it is also possible to use both),
which must be specified at build time using a special CMake parameter (see below).
You can download them from their respective websites, then build the libraries
(libtokyocabinet and/or libcouchbase) and install the binaries and headers in
your system (the headers should be put into a
directory respectively). If you want to use other databases, you can use the
provided drivers as blueprints to write your own.
Audio data is read from files or audio devices using the
AudioSource class, which
by default looks for the FFmpeg audio decoder on the host system. It needs to be
located somewhere accessible from the PATH environment variable in order for it
to be found and used. Should you want to use a different audio decoder, you will
need to modify the
The examples use the Boost library for several processing tasks (filesystem access, threading, events handling, etc.) so you need to install that too. At least version 1.55 is required. The only needed modules are thread and filesystem (and their dependencies), so you can build just those two if you don’t want to build the whole thing.
Optionally, if you wish to extract metadata from the audio files, to be used in the identification process, get the TagLib library as well.
The Linux and Windows examples can be built as part of the main library build process by indicating so in the build script’s command line, as follows
$ ./build WITH_EXAMPLES=ON
By default, the programs will use the Tokyo Cabinet database through the
TCDataStore driver. It is possible to change this option and use the Couchbase
database by specifying the
DATASTORE_T parameter as follows
$ ./build WITH_EXAMPLES=ON DATASTORE_T=CBDataStore
If you want to build with ID3 tag support to handle the metadata then include
$ ./build WITH_EXAMPLES=ON WITH_ID3=ON
The Android examples can be found in the
In order for the apps to successfully build there are some things to
configure beforehand according to your environment.
The Tokyo Cabinet headers (tc*.h) must be installed into a
tcabinetfolder, along with the Boost headers, somewhere in your system.
Build all the required native libraries and put them into a
/libfolder in the root directory using a path with the following scheme
buildscript will do just that for the Audioneex libraries. External ones, you need to build them yourself and then copy them into the above mentioned directory. Patched sources for the default libs for Android can be found here. Just run the
./android-configurescript located in the root directory of these packages instead of the standard ./configure.
Locate the CMake script file in the
app/src/main/cppdirectory of the Studio projects and set the include paths in the User Config section to the folders where you installed the headers at step 1. If other include paths are required just put them there.
The examples use the CMake build system, so you need to have it installed and configured properly. However, using the CMake installed system-wide requires some extra steps (like installing additional dependencies). The easiest way is to get the one shipped with Android Studio through the SDK Manager, which already includes the required components (version 3.6 is recommended at the time of this writing as the more recent version doesn’t work). You will also have to set the Android Studio SDK and NDK paths to the directories where they’re installed in File->Project Structure. After that, everything should be set to go. An internet connection is likely to be needed for Gradle to download some dependencies. The projects have been set to build for armeabi-v7a architectures by default. If you need something different then modify the filter in the app’s Gradle script.
About the Android demo apps¶
This app is meant to verify that the main functionality of the engine work properly on the target device. Just run it and click the button to start the test. If everything works well you will see a success message.
This app demonstrates how the engine can be used to perform over-the-air
recognitions. In order for the app to work you will need to put a fingerprint
database (*.idx, *.qfp and *.met files) into the
assets folder. The database
can be created using the command line programs. It can also be done
programmatically in the app, but you will need to write the code for that.