How to build it¶
The code has been extensively tested on Linux, Windows and Android platforms, so here we’re outlining the steps needed to build the binaries on such platforms. However, it may as well build on others not officially mentioned here. Generally, any platform with a modern C++ compiler (version 11 or higher) and a decent POSIX compatibility should be fine.
The engine itself only needs the following dependencies to build and run
Tokyo Cabinet 1.4 / Couchbase 5.1
To build the full package, including the examples, you need to add the following optional components
The code has been developed mostly using the below mentioned tools, but anything more recent should also work fine, so these are considered the minimum requirements
GCC 6, CLang 5, MSC 19.1
Android NDK r19
TagLib and FFmpeg are only needed for the example programs and can be replaced with something similar, but doing so will require changes to the code.
About the database¶
Audioneex needs a database to store the fingerprints and is designed to be database-neutral, so technically it can be used with any database. However, using databases other than the ones supported out of the box requires writing the drivers. The default databases are Tokyo Cabinet and Couchbase. The former is an embedded/in-process database (suitable for mobile/embedded apps), while the latter is a client/server type.
The project uses the CMake build system on all supported platforms. The steps to follow are pretty much the same for all platforms, aside for few specific tweaks that may occur.
1. Set up the build environment
Install Boost. The library requires the header-only part, but the examples will need some compiled modules (thread, filesystem, regex and their dependencies).
Get the FFTSS library, compile it in static mode and install it somewhere in your system (remember to compile with the
-fPICflag on Linux otherwise linking errors will occur).
Get the Tokyo Cabinet library (for Windows there is a port from the EJDB project here). Alternatively, you can use the Couchbase database, which has a nice and free “community edition” (requires building their libcouchbase driver).
Get the TagLib library for ID3 tag support (used by the examples to extract metadata from audio files).
Get the FFmpeg executable and install it in a location visible from
PATH(used by the examples to decode and read audio).
2. Set include and library paths
This step is not mandatory but it will most likely be necessary since these paths are system-dependent. You can set them in the User Config section of the main CMake build script (CMakeLists.txt) located in the root directory.
After cloning (or downloading) the code, open a shell and execute the following command from the root directory
$ ./build [options]
[options] is one or more of the following command line parameters in
TARGET = linux | windows | android ARCH = x32 | x64 (Linux/Windows) armeabi-v7a | arm64-v8a | x86 | x86_64 (Android) API = API version number (Android only) BINARY_TYPE = dynamic | static BUILD_MODE = debug | release DATASTORE_T = TCDataStore | CBDataStore WITH_EXAMPLES = ON | OFF WITH_ID3 = ON | OFF (for the examples only) WITH_TESTS = ON | OFF (for project developers only)
By default, if no options are passed on, the build targets Linux (or Windows)
x64 and produces the dynamic library only, in release mode. The final libraries
will be put in a
/lib folder in the root directory.
To build for Android, specify the target as follows
$ ./build TARGET=android [options]
If no other paramaters are given, then the build defaults to the armeabi-v7a architecture and the latest API level.
The parameters for the build script are case-sensitive.
Naturally, first you will have to build the required external libraries mentioned in the prerequisites for the specific Android platforms you’re targeting. Patched source code for the libraries that compile on Android straight away can be downloaded from here. Just unpack them somewhere and run
$ ./android-configure <arch> <api> [config_params] $ make
from within the respective directories, where
<arch> <api> are the same
as in the
build script and
[config_params] are library-specific
configuration parameters. Please have a look at the script for more details.