Chatterbox-Turbo default settings.

[sruckh]

On the Chatterbox Turbo demo site on hugging face, I uploaded a 27 second audio (.ogg) reference, with a small paragraph using a few Para lingual tags, using the default settings. Oddly, without hitting the generate button it started playing audio which sounded great. When I pressed generate and downloaded the audio it sounded absolutely nothing like the audio that was playing earlier. The output is just gibberish, the audio does sound like the reference audio, but the content is just a garbled mess of sounds and partial words. This is exactly the same experience I am having with the API I have implemented using the Chatterbox Turbo model. It is unclear if it is the default parameters or something else, but I am not getting anywhere close to he same results as those published on the example page. As I am getting the same results from the huggingface spaces page, and my own implementation, I am wondering how the published results were created as my experience is nothing close to the published audio examples.

Thank You.