ANTHONIO PETTIT -  LE&C ENGINEER (2018.3.30)

The LE team has been piloting a gesture-based automation schema to validate the thousands of K2 audio sample transcriptions using a third-party application called StrokeIt1.

Because the Utterance Audio web interface contains consistently spaced controls and tab order, it is possible to design macros to focus the browser caret, start audio playback, and complete relevant survey questions with a minimal learning curve. Once memorized, the gestures significantly increase validation speed and accuracy.

Macros are created in the StrokeIt Command Editor user interface, and assigned a gesture to activate the sequence of command.

The gestures can be easily customized to the current task and adapted to any page layout if the tab order remains consistent between pages.

Sets of gestures and commands can be grouped by each task and exported as a .SXP file which can be shared and activated or deactivated as needed.

Once configured, macros are activated by holding the right mouse button and drawing the corresponding stroke listed on the task’s gesture map. Each gesture triggers a specific sequence of keystrokes which select relevant validation tags, and advance to the next audio clip automatically.

Each page requires an initial Focus gesture, which moves the browser caret to the first audio clip and starts playback. After hearing the clip, the user determines the accuracy of each transcription. Specific gestures mark the transcription as 100% accurate or specifies a minor or significant difference, or that the audio is clipped or unintelligible. For minor or significant differences, the caret moves to a text field where a better transcription can be input. Supplemental gestures allow the clip to be replay without losing the text insertion position. Clipped or unintelligible clips are marked and advanced to the next audio file.

1 http://www.tcbmi.com/strokeit/

2

StrokeIt development appears to have ceased several years ago, but the application is still available from the developer’s site. On Windows 7/10 the application must be run with elevated privileges, with Compatibility Mode set to Windows XP SP3.

ADVANCED MAPPING EXAMPLES

The highly flexible gesture engine allows for quite extensive sets of key commands and strokes to be created and executed quickly by drawing strokes in a specific order. Below are examples of gesture mapping automation for various Crowdflower page layouts and task requirements.

K2 DATA-3326.sxp

K2 Intellible and Clipped Gestures.sxp

The use of gestures has turned the validation tasks into a satisfying and fun-to-do task, and was adopted quickly by those who were able to configure it properly.

OTHER USES

StrokeIt can be configured globally or on a per-app (or even per window class or title) basis. Another early use of this tool was to insert an array of frequently used tagging facets in Excel. Tags can be rapidly inserted using gestures to trigger keystrokes which would otherwise require manual entry.

In summary, StrokeIt is a powerful and extensible tool which has the potential to increase efficiency and accuracy in a variety of situations and tasks.