Q: Quality of text to speech conversion
So here's a fun little question:
What controls or mechanisms are in place to determine how a word is pronounced?
Using your "Gordon Ramsay" AI Voice as an example: https://www.lazybird.app/gordon-ramsay-ai-voice
"isn't" is pronounced "i-s-n-t"
It pronounces "It's" just fine a few words prior, but there should be some degree of confidence that basic contractions are pronounced correctly so that you can focus on more nuanced, niche words that would require a fine touch - but again, is there any way to address those either? For example, "Gyokeres" is absolutely butchered in the free front-end tool. Just one example of many I'm sure users can produce.
Ellis_Lazybird
Jan 22, 2025A: Hey, that's a really bad hiccup from Gordon :).
But no worries, in the full app at https://studio.lazybird.app you have the ability to modify the pronunciation as you like. In this case, just select the word "isn't" and look on the left side, you'll see a button to Modify the pronunciation, enter "ɪzənt" in the IPA code and apply. It will pronounce this correctly.