Has something changed? I've wasted all my credits (20k+) and the voice isn't what it used to be.

[serene phoenix]

Hi Oscar, totally fine to tag me.

There has been no change to the models for the last few months. The latest change was when we added the Turbo v2 model which was not very recently anymore, and to the other models, there has been no change for quite a while. Are you sure you are using the exact same settings, model, and voice?

Definitively, please do! I would be happy to listen to that if you can download some examples from your history to show the difference.

That was release a while ago with the v2 models. So maybe 6 months ago. I don't exactly remember when those models went into alpha.

Thank you!

I agree. It is quite amazing. This is something I'm testing out now and I'm blow away.

This is with non-ideal samples as well.

[serene phoenix]

It is really mindblowing how good it is. If you are not getting these kinds of results (the above are fantastic so might not be viable for everyone), but you should get farily good results, there might be something with your samples or settings.

[serene phoenix]

Update: This has been resolved, and if anyone else reading this, it was an issue with the settings/models used. The difference was between using the v2 model instead of the v1 model that had been used for the old generations.

[pallid shore]

Hi, I would have expect than v2 model to be better though?

[serene phoenix]

Hi! Well, it is subjective so what you might like, other people might not like. Generally, the v2 models are "better" at least more accurate (the English v1 is still the most stable model, and the multilingual v1 is the least stable), but some people like the sound and deliver of each model differently.

[pallid shore]

In my case, I am French, so v2 is much better. My only concern is related to Pauses. after some pause, I get a Spanish accent for my French paragraph. I am using starter plan with instant voice.