rsxdalv
Bark voice clone multiple audio inputs?
It seems this isn't possible? What would be an ideal audio file length for Bark voice cloning if it can only accept a single input? I guess this might be a reason to use Tortoise instead. Usually the larger the dataset, the more accurate the reproduction.
Bark voice clone is a lot more like stable diffusion with hit and miss. There are some guides and explanations, but generally 6-10 seconds should be good.
Tortoise can do a lot better voice reproduction if that's your specific goal.
On Mon, Jan 15, 2024, 12:57 AM MysticDaedra @.***> wrote:
It seems this isn't possible? What would be an ideal audio file length for Bark voice cloning if it can only accept a single input? I guess this might be a reason to use Tortoise instead. Usually the larger the dataset, the more accurate the reproduction.
— Reply to this email directly, view it on GitHub https://github.com/rsxdalv/tts-generation-webui/issues/254, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTRXI2OLR7SHUEIAOIDOBDYORPHFAVCNFSM6AAAAABB2NREZKVHI2DSMVQWIX3LMV43ASLTON2WKOZSGA4DAOJTGQYTEMY . You are receiving this because you are subscribed to this thread.Message ID: @.***>
The turtle can reproduce the voice much better - I agree 100%, it's a pity that the language possibilities are so limited, I'm looking for a Polish model :) or a description of the possibility of training your own language model at home - a simple model
The turtle can reproduce the voice much better - I agree 100%, it's a pity that the language possibilities are so limited, I'm looking for a Polish model :) or a description of the possibility of training your own language model at home - a simple model
XTTS is very similar to tortoise but has polish support built in. There's a plugin that I will continue improving for running XTTS.