Wekinator

The Wekinator is an open source software designed for real-time, interactive machine learning. According to the vendor, it allows users to build interactive systems by demonstrating human actions and computer responses, eliminating the need for coding. The Wekinator is suitable for businesses of various sizes and caters to professionals in industries such as music, art, game development, computer vision research, audio engineering, and interactive media design.

Key Features

Real-time, interactive machine learning: According to the vendor, users can create new interactive systems by demonstrating human actions and computer responses without the need for programming code. This feature enables the creation of musical instruments, gestural game controllers, computer vision or computer listening systems, and more.

Creation of new musical instruments: The Wekinator allows users to create mappings between gestures and computer sounds. It provides the capability to control a drum machine using a webcam and play Ableton using a Kinect.

Creation of gesturally-controlled animations and games: Users can control interactive visual environments or game engines like Unity using gestures sensed from webcams, Kinect, Arduino, and other compatible devices.

Creation of systems for gesture analysis and feedback: The Wekinator enables users to build classifiers that detect user gestures. According to the vendor, these classifiers can be used to control the computer or provide feedback on performance.

Creation of real-time music information retrieval and audio analysis systems: The Wekinator can detect instrument, genre, pitch, rhythm, and other audio attributes from microphone input. Users can utilize this information to control computer audio, visuals, and more.

Creation of other interactive systems: According to the vendor, the Wekinator supports the creation of any interactive system where the computer responds in real-time to human actions. It is compatible with systems that can output OSC (OpenSoundControl) and can be controlled by OSC.

OSC Message Control: The Wekinator can be controlled via OSC messages, allowing users to send commands to control its behavior. Users can start/stop recording examples, train on current examples, start/stop running, and enable/disable model running and recording using OSC messages.

Dynamic Time Warping (DTW): The Wekinator supports dynamic time warping, a technique used for comparing and matching time series data. Users can record examples for DTW gestures, train them, and use them for real-time classification and recognition.

Model Management: The Wekinator provides a user-friendly GUI for managing models. Users can enable/disable running and recording of models, train or cancel training on current examples, and delete all examples or examples for specific output/gesture classes.

Input and Output Configuration: Users can set the names of inputs and outputs in the Wekinator. The number of inputs and outputs can be adjusted to match the desired configuration. Users can also select which inputs are connected to a particular output, allowing for flexible feature selection.