After giving this new model a try, I am loving the speed and accuracy and how realtime it is. However, here are some issues I noticed:
1st issue:
About every 30-45 seconds, the voice randomly gets super excited and almost starts yelling, then calms down. The original speaker themselves was speaking calmly and not in any exciting manner. Randomly the voice can get quiet, fuzz out, then abruptly go back to normal. Also want to mention there is no consistancy in voice tone overtime and seems like the speaker voice randomly switches here and there.
2nd issue:
The text token output randomly does not properly space words. For example:
“But the time camewhen they saw something,”
I can see in my logs that this issue stems from the api and nothing to do on my client end.
