I've found that Amazon Polly NTTS provides a much more natural and accurate voice than Google TTS inside of Genesys. I don't know if you can do what you are asking, but you can manipulate the playback quite a bit using the markup language. In the end, I've found that using the Polly console to test playback and tweak it via the console, then downloading it to a wav file to be best. I put those wav files into the IVR whenever we have static prompts / statements, then use the Polly TTS engine when it needs to playback numbers, or any other dynamic statements. This way, there are a lot less API calls to the TTS engine, the IVR runs more reliably, and you get a unified voice. There are caveats with both Google TTS and Amazon Polly TTS, like the issue with In-Queue flows and agent whispers not working.
------------------------------
Ryan Cheesman
Senior Manager, IT Integration Services
Tandem Diabetes Care Inc. | positively different
------------------------------