Avoid Watson Unity SDK for speech recognition

[Zane]

I’ve got a very simple Unity “game” that I work on for academic purposes. Only one scene, deployed on iOS only, it’s basically just a chatbot. We deploy this thing through an enterprise distribution platform, and I have to hand off the file to an IT team to get it out to all of the devices, so it’s a little more of a PITA to deploy than if I were going straight to the App Store. We’re trying to make the dialog interaction as natural as possible, so we switched from typing to speaking, and we decided to use Watson to handle the speech recognition and text-to-speech generation.

I implemented the spoken interface almost a year ago. Since then, the Watson development team have pushed breaking changes to their service and/or API at least twice. I’m currently spending the day upgrading my API version and figuring out how to modify my existing code, since it no longer works with the new API. Maybe I’m being unreasonable, but it seems to me that backward compatibility shouldn’t be a radical concept for something people pay for.

Anyway, if you depend on speech recognition in your game, save yourself some trouble and try something else. I haven’t used Google’s or Amazon’s speech-to-text services, so I don’t know if they’re any better, but I’m tired of the hand-holding I have to do with Watson. Just thought I’d give everyone a heads up!

[Author]

Posts