@xezpeleta
Thanks for sharing the ideas!
1. shell prompt works (I have been using Python3)
2. [surprise] actually changes the emotion!
Once, I was able to get [laughter] to sound like a real laugh.
Command used for the following audio:
omnivoice-infer --model k2-fsa/OmniVoice --text "[surprise] Can you actually generate audio with a shell prompt?!" --language en --output demo_tag3.wav

