Right now the AudioToAudioTranslateProvider class is registered as a taskprocessing provider no matter what.
Inside this class, we throw in
- getOptionalInputShape
- getOptionalInputShapeEnumValues
- getOptionalInputShapeDefaults
if no STT (core:text2speech) provider is found (because audio translation depends on STT).
This produces a lot of level 3 logs.
We could improve this by either:
- Finding a way to only register
AudioToAudioTranslateProvider if STT (and the other task types we depend on) are found. Not sure it's possible to check this in the Application::register context.
- Find a way to silently fail, meaning the taskprocessing manager would not produce some logs in this case.
Right now the AudioToAudioTranslateProvider class is registered as a taskprocessing provider no matter what.
Inside this class, we throw in
if no STT (
core:text2speech) provider is found (because audio translation depends on STT).This produces a lot of level 3 logs.
We could improve this by either:
AudioToAudioTranslateProviderif STT (and the other task types we depend on) are found. Not sure it's possible to check this in theApplication::registercontext.