Lowwi - Lightweight openwakeword implementation written in C++.

Wakeword detector based on OpenWakeWord-cpp but faster, cleaner and more lightweight.

• Plain C/C++ implementation with minimal dependencies (ONNXRuntime)
• Runs on ARM as well (Tested on RPI 3,4 and 5)
• Written with performance in mind, < 3% CPU utilization on x86/64 CPUs

API Features

This library offers support for:

• Single-model wakeword detection
• Multi-model wakeword detection

Single-model wakeword detection

This is some example code for wakeword detection:

...
 /* Create new Lowwi runtime */
CLFML::LOWWI::Lowwi ww_runtime;

/* Create new wakeword */
CLFML::LOWWI::Lowwi_word_t ww;
ww.cbfunc = wakeword_callback;
ww.model_path = "models/example_wakewords/hey_mycroft.onnx";
ww.phrase = "Hey Mycroft";
    
/* Add wakeword to ww-runtime */
ww_runtime.add_wakeword(ww);

std::vector<float> audio_samples;
...

while(1) {
    ww_runtime.run(audio_samples);
}

Multi-model wakeword detection

Just create two or more Lowwi_word_t's:

...
/* Create new wakeword */
CLFML::LOWWI::Lowwi_word_t ww;
ww.cbfunc = wakeword_callback;
ww.model_path = "models/example_wakewords/hey_mycroft.onnx";
ww.phrase = "Hey Mycroft";
    
/* Add wakeword to ww-runtime */
ww_runtime.add_wakeword(ww);

/* Create new wakeword */
CLFML::LOWWI::Lowwi_word_t ww2;
ww2.cbfunc = wakeword_callback;
ww2.model_path = "models/example_wakewords/hey_jarvis.onnx";
ww2.phrase = "Hey Jarvis";
    
/* Add wakeword to ww-runtime */
ww_runtime.add_wakeword(ww2);

Example code

For a full example showcasing the API functions see the example code in example/LOWWI_demo_mic/demo.cpp.

Building with CMake

Before using this library you will need the following packages installed:

• Working C++ compiler (GCC, Clang, MSVC (2017 or Higher))
• CMake
• Ninja (Optional, but preferred)
• SDL (when using Lowwi mic example)

Running the examples (CPU)

1. Clone this repo
2. Run:

cmake . -B build -G Ninja

3. Let CMake generate and run:

cd build && ninja

4. After building you can run (linux & mac):

./LOWWI_demo

or (if using windows)

LOWWI_demo.exe

Usage with ROS Pixi (Highly recommeneded with ROS2)

See the wiki page!

Usage with ROS prebuilt .deb package

See the wiki page!

Using it in your project as library

Add this to your top-level CMakeLists file:

include(FetchContent)

FetchContent_Declare(
    Lowwi
    GIT_REPOSITORY https://github.com/CLFML/lowwi
    GIT_TAG main
    SOURCE_DIR ${CMAKE_CURRENT_LIST_DIR}/lib/Lowwi
)
FetchContent_MakeAvailable(Lowwi)
...
target_link_libraries(YOUR_EXECUTABLE CLFML::Lowwi)

Or manually clone this repo and add the library to your project using:

add_subdirectory(lowwi)
...
target_link_libraries(YOUR_EXECUTABLE CLFML::Lowwi)

Aditional documentation

See our wiki...

Todo

• Add language bindings for Python, C# and Java
• Add support for MakeFiles, Bazel and Conan
• Add Unit-tests
• Add TPU support
• Add automatic wakeword training
• Merge melspectrogram & embedding model?

License

This work is licensed under the Apache 2.0 license.