Kinect for Windows SDK beta

What the beta was

On 16 June 2011 Microsoft released a beta version of the Kinect for Windows software development kit, a free download intended for non-commercial use.^[1] A software development kit (SDK) is the bundle of drivers, programming interfaces and documentation that lets someone write programs for a piece of hardware without starting from the raw electronics. This one had a specific aim: to make the Kinect sensor's human motion tracking, voice recognition and depth sensing usable on ordinary Windows PCs.^[2]

It was a step forward for natural user interfaces — computer input built around voice, gesture and body movement rather than the keyboard and mouse.^[2] Researchers and enthusiasts who wanted to build such interfaces could now work through supported Microsoft tooling and documentation.^[1]

What was in the kit

The kit contained four parts: drivers; rich APIs (application programming interfaces) for the raw sensor streams and for natural user interfaces; installer documents; and additional reference resources.^[1] Raw sensor streams means a developer can read the sensor's data directly instead of only receiving a finished answer such as "a hand is here", while the natural user interface layer works at the higher level of gestures, speech and body position.

Developers could write in C++, C# or Visual Basic, using Visual Studio 2010.^[1] The complete installer was under 100 MB and required no complex configuration.^[1] A small download, languages many developers already knew and a familiar development environment kept the barrier low for a university lab or a single enthusiast.

Audio and speech

Audio received separate attention. The beta included acoustic noise suppression, echo cancellation and beam formation, and it was integrated with the Windows speech recognition API.^[1] Echo cancellation matters when a microphone can also hear the loudspeakers in the room; beam formation combines the signals of several microphones so that sound arriving from one direction is favoured and the rest is suppressed. Paired with the Windows speech API, these pieces let an application accept spoken commands without its author writing a speech recogniser from scratch.

What it made possible

The SDK enabled human motion tracking, voice recognition and depth sensing on PCs.^[2] In practice that meant a program could work out where a person's body was in the room, how far away objects were, and what was being said — the three signals from which gesture and voice interfaces are built. Because the kit was free for non-commercial applications, the earliest projects came from academic researchers and enthusiasts rather than from product teams.^[2]

Where it sits

The beta was explicitly a first step: it was free for the development of non-commercial applications, and a commercial version was expected later.^[2] That framing set the pattern of an academic and hobbyist phase first, with the sensor's use on Windows opened up in stages. By packaging drivers, interfaces, documentation and speech integration into a download of under 100 MB, Microsoft made the sensor's depth and audio data reachable from ordinary desktop code.^[1]