luketurnerdev
commented
1 year ago Update: I have been trying the voiceSDK package v53 (https://developer.oculus.com/downloads/package/oculus-voice-sdk/) instead of wit-unity, and still cannot get transcriptions. This is in the demo scene:
I can get an output in the 'last transcription' section, and the callback fires (as seen in debug logs), but the string value is not attached. When using the voiceSDK in my actual project I am not even able to get this callback to fire.
Just not sure if this is an issue on wit's end or mine. Appreciate any feedback. @yolanother may I request you look at this please?
Hi guys,
I have been building a project using this package. Since last night (4/7/23) I am unable to receive a partial or full transcription callback from Wit. App ID: 1004534554324678. Unity ver 2022.1.4f1.
My app setup is as follows:
I have a Wit Gameobject in my scene with a Wit script attached. It has a bunch of callbacks, including OnPartial and FullTranscription. I use a script to handle these and display them in game. They are no longer firing.
Things I have tried:
1) Ensuring the transcription handling script compiles, and is attached to the event handler in the inspector for both callbacks:
2) Using the understanding viewer - it starts in a 'listening' state but doesn't listen to any of my mic inputs. Hitting Deactivate does nothing. Hitting abort and typing a request yields a response (no entities etc are setup ):
3) I have created a Meta account as I can see that you merged from Facebook accounts recently. I created a fresh app and set it up in the wit configuration. (The app id found above).
4) Updating the package gives :
I resolved this by removing the package and re-adding it via the git URL, but then it started with these two errors:
which I was able to fix by removing the 'override' keyword, and using UnityEngine.Editor instead of 'Editor' as the inherited class of TTSSpeakerInspector. (It still didn't work after this)
5) Debugging various request states; the following callbacks fire successfully:
-OnMicVolumeChanged fires, but always has a value of 0 for the volume.
6) Verifying my mic works, and trying a different mic.
7) Getting a colleague to verify he is having the same issue.
8) Going back to a previous commit of my project and trying it (where voice should have been working) to no avail.
Sorry for the long winded post. I understand this is an experimental package and appreciate your work and support. I want to ultimately alter the project to simply use VoiceSDK instead of this package, can you reccomend some instructions on how to achieve that?
Given that I only need a few callbacks (partial/full transcription, onstart, onstopped, onmicdatasent), hopefully the voiceSDK should be sufficient and the transition not too painful?
Thanks in advance! Luke