AmazonLexRuntimeV2Client.RecognizeUtteranceAsync RecognizeUtteranceResponse field is compressed with gzip and then base64 encoded.

I am trying to link unity3d with Lex using .net sdk. I am able to connect to the Lex bot using RecognizeTextAsync. However, I am not able to use RecognizeUtteranceAsync because the respond has many fields that are gzip and base64 encoded. I have tried many suggestion from stackoverflow to unzip and decode the response. None of them work. Does anyone have sample or tutorial that work with RecognizeUtteranceAsync? Thank you.