Voice recognition is a driving element in our voice activated project. We had the choice of using voice control either through mobile input via Android phone (AMR App) or a voice recognition component itself. In such projects, things usually work through trial and error hence we decided to test out the Voice Recognition V3.
How the VR works
The module requires you to speak the commands several times to ‘train’ the device to pick up your speech before registering it. We expected it to be fairly simple.
Test and outcome
For a trial, we recorded the basic ‘on’ and ‘off’ voice command using my voice. We then tried to use each of our voices to speak into the mic to test if it picks up different voices. To our horror (and frustration), it was already difficult to pick up my voice, which was what was used to record and train the commands.
Therefore, we have decided to stick to the AMR voice recognition as it detects phonemes of the word, in contrast to the VR module that picks up the way a person speaks. This could be rather limiting for the VR module.
Also, we find that throughout this process of the project, it’s also helping us to enunciate our speech better. This could be a possible alternative use for this project.