#what has changed in the multiligual V2 model over the last month? there are now lots of odd breakup
1 messages · Page 1 of 1 (latest)
The Multilingual v2 model is designed to provide high stability and accuracy across 29 languages, but occasional issues like inconsistencies or glitches can still occur. While there is no specific mention in the provided content about recent changes to the Multilingual v2 model in the last month, it’s worth noting that AI models can sometimes exhibit variability due to updates or other factors.
To address the odd breakups in speech you're experiencing, here are a few suggestions:
-
Adjust Settings: Ensure that the stability is set to 100% and style is set to 0% for more consistent results, especially if the input audio is expressive or dynamic.
-
Use High-Quality Inputs: The accuracy of the model heavily depends on the quality of the input samples. Consistent, high-quality voice samples can help reduce issues.
-
Regenerate Problematic Sections: If the issue persists, try regenerating the specific sections where the breakups occur, as this can often resolve rare glitches.
If the problem continues and you suspect a change in the model's behavior, you may want to open a ticket via the ticket form to report the issue and get further assistance.
i've tried that but that doesn't stop the slight glitches that happen now