#Voices Drastically Change Tones, Making Them Unusable
1 messages · Page 1 of 1 (latest)
Hey Trolleater,
Voice stability can indeed vary depending on the voice you're using, particularly based on the samples used to create it. If you're using a professional voice clone, please ensure you're use the recommended model for the voice.
For our default voices, they should be very stable when used with the correct model. If you're still encountering issues, we'd be happy to look into it further for you, but would need more information.
I was using a professional voice and it was on the suggested model.
The final result isn't usable though.
Then it is very likely the creator of that voice did not use optimal samples.
We have a guide and the samples should be extremely consistent to get a good, consistent output, but that is more of a guideline. You can create a clone without following those and some of those are a great, albeit a bit more unstable usually.
For voices that are more dynamic and less consistent, you should be using Projects (or Request Stitching if using the API) to ensure consistency between separate paragraphs. Maybe you're already doing this, in which case, it's likely the voice itself which is the problem.
Voiceover Studio and Dubbing Studio may also work for improved consistency, I'm not sure. But Projects definitely uses previous generations to influence the next to maintain consistency, which separate paragraphs sent through Speech Synthesis will not.
I did use Projects, and from listening to the output and seeing several people on Discord complaining that this started happening to them yesterday, I'd assume it's on your end.
It does it at the beginning of every 3rd-4th paragraph, at the very beginning. It's never in the middle of a sentence, or in the middle of a paragraph...it's always the beginning. About 80% of my project is fine, but that's of no use to me...and I don't want to be paying for something I can't use, especially when it has a 3x multiplier.
I'm really sorry to hear about your experience. Would it be possible to share some more details?
- Voice used
- Model
- Settings
- Audio examples
- Text used
Having a bit more context would make it easier to see if there is something we can suggest that might help minimize these issues for you. In general, it is highly dependent on the voice.
It seems like it's your system. I found that if I do 1,000 characters at a time, it seems fine. I tried another voice and the same thing started happening after about 2+ mins in. It's frustrating wasting characters trying to figure out the quirks of your system.
Initially I used Jake's voice and my settings were...
Then I tried Wade's voice and like I mentioned, after about 2 mins his voice "bogs" out.