I recently inherited a project for HoloLens 2 using Unity 2019.4.11f and an old version of MRTK. My job on the project has been integrating text-to-speech, which I have done using Microsoft Cognitive Services Speech. So far, things work great in the editor, but when the app is deployed to the HoloLens, there is a significant lag during scene transitions (on the order of 10-20 seconds, or sometimes longer). The app also blacks out frequently during this loading period. All builds I have tried so far have been debug builds in Visual Studio. This issue was not present in the last build before I worked on this project, but I do not understand why or if my changes caused this issue.
So far, I am considering alternatives to Microsoft Cognitive Services for text-to-speech, but I would like to explore other solutions before making that change.
Does anyone have guidance on achieving faster scene transitions in Unity? My understanding is that SceneManager.LoadScene() is operating without a delay, so I want to understand why this bug is occurring.
I also fear that upgrading the version of Unity will lead to dependencies like MRTK breaking, which I have experienced before. Any guidance would be much appreciated.