Voice data collection in GlobIQ

By November 10, 2020 Artificial Intelligence

Artificial Intelligence is becoming an inherent part of our reality. Using it in applications improves work efficiency and that saves money. We use AI in the GlobIQ project as a part of GeoTask, the mobile field service management application that includes forms for collecting data in the field. Learn more about how AI mechanisms implemented in GeoTask make field data collection easier and more efficient.

AI has been used in applications with voice commands for a long time now. When we worked on using voice in the mobile solutions, we searched for a way to effectively enter data in forms. So, why voice commands? After years of working on geospatial projects, we’ve learned that using traditional, manual forms during fieldwork is troublesome.

The most common problems with manual data entry are:

  • a field worker has to scroll through the screen to find the right attribute
  • long drop-down lists have to be scrolled through to find the right device type
  • it’s hard to tap the right spot on the screen, especially when keyboard keys are small

All of this makes the fieldwork workers unmotivated to use typical form applications so using voice seems to be the best solution. The typical voice assistant isn’t ready for capturing many different types of data at once. This is why we developed it to solve this problem and improve using forms during fieldwork.

How does it work?

The device screen shows a form to fill out and a microphone icon.
The user taps the microphone icon or says the wake-up command to initiate the voice assistant.
The user can say all the words for filling out many fields in one sentence, following the <field_name> <field_value> protocol.

The application is fully interactive. The user’s speech is analyzed as they speak. The application scrolls the screen, highlights fields mentioned by the user, and fills them out with dictated text on an ongoing basis. This way, the user gets a visual confirmation of the text they just dictated to the app.

Additionally, we implemented a synonym function for field names and field groups in the forms so that the user can say them in many ways. We also developed a way to dictate the data in order of their appearance in the form. We wrote a module for text processing for collecting the information and filling out the forms.

The work is still going and in the next stage, we will focus on improving the application’s GUI.
If you’re interested in more details or a live showcase, let us know!