Integration of speech-driven data entry
looking for funding
Micah Duckworth
I use Qfield on an android tablet that has speech to text as an option on the default keyboard as well as handwriting recognition using a stylus. It works pretty well for data entry. I can say oak when in the search box of a drop down selector and it selects for me.
S
Sus
I feel your need, but how would you work that out with both hands free?
Is local AI with Whisper + AI taking over QField editing already within reach?
To bridge that gap I have made a small Android app that handles like a simple voice recorder with the volume buttons, tracks gps and exports a point layer with associated audio files.
The app is still rough and limited, but already offloads much hassle from the field to the comfort of a dry desk with QGis and warm fingers.
I would love to see this integrated in QField.
Z
Zsanett Medvei
Sus: Thank you, I've removed that part. :)
Jean Cyane
That sounds great! I have been looking for a way to implement something like this in my line of work as well.
However, it seems insanely difficult to implement in QField.
In my case, I collect environmental and ecological data. I have basic free-writing text boxes, but I also have text boxes with constraints, combo boxes, lists, intervals, etc. I would also need to be able to save scientific names (Latin).
Another issue : it has to work offline... I don’t know if a fully functional text-to-speech system could work offline.
Being able to add just one entity on the map and complete all the required fields would be quite complex.
This feature would either have to adapt to every situation (which I don’t think is possible), or it would have to be configured through QFieldSync (or somewhere else) and tailored to each project.
A small first step for me would be to allow the user to dictate which field to focus on for completion (e.g., say “number” and the “number” field is selected on the form, ready to be filled in). But not very useful.
Z
Zsanett Medvei
marked this post as
looking for funding