Foremost, some fixes to pitch estimation. It should prevent a lot of the buzzing that was happening before.
NEW MODEL! This one works better for voice cloning. It doesn't work for all voices, but some sound quite decent.
File conversion has had some fixes.
Two options that can further help with pitch:
-Smooth Pitch: can further reduce buzzing, gets rid of sharp changes in pitch
-Continuous Pitch: helps with very rare pitch errors, won't be noticeable usually
...
Some big changes, mostly to the UI, but also some model adjustments and added features.
-New UI. Darker, sleeker, probably going to change for the next release. Red is a very aggressive color. Once I find a color scheme I like then I'll swap over. You can change the UI color in the settings yourself if you want to try different colors.
-New models A and B. B is chosen by default and the pros and cons are as follows.
B: slightly more 'robust' and better at handling unseen words...
First, a new experimental build is available to those interested (from the Google Drive link above). This one has improved pitch detection which adds additional quality to the voices (see below for an example of the quality to expect). I would appreciate any feedback offered, particularly with how the pitch and quality in this one compare to the previous ones. It still has some audible artifacts which are being investigated.
Second, I am working on a complete rework of NyVox which will...
This is a quick update to the last version I posted yesterday. There was a memory leak that caused NyVox to use increasingly more ram, but it has been patched. Thank you Fushi in the Discord for pointing it out!
Some instructions on how to use it for singing that I left out. First check Details, then either uncheck Dynamic Pitch Mean OR check Manual Pitch Values and adjust the slider to your voice mean.
I plan on adding a few Community Voices to this to work on the fine-tuning p...
This is an experimental model for those curious about some of the testing I have been up to. Progress is not always linear and often times you have to try new and different things to get where you want. This is absolutely not a final model or even one I consider worthy of an update, but it has some neat features.
It handles pitch much better and even allows for singing. Additionally, some of the 'wobbliness' has been reduced. However for this model, it comes at the sizable cost of add...
This is not the final version, this is mostly just an update to address some fixes and test some things out. It's been a while since I posted an update and I want to keep everyone informed even if things are not done yet. In fact, this updated version may be worse in some ways, so do give it a test first.
The download is now an installer, here is the link: https://nyvoxdata.com/v1/NyVoxSetup.msi
(Pl...
This is an early test version of NyVox. A Steam version is still in the works as well. Manually requesting and distributing Steam keys was not an ideal solution so hopefully, this will work better.
This update includes several quality improvements along with Voice Mixing which lets you customize a voice to your liking!
Wav file conversion is not included in this release.
2022-07-27 20:29:56 +0000 UTC View Post
NyVox 2.0 is up for testing on Steam! If you are a subscriber and have not received a key, write me a message and I will send you one! Be aware there is some functionality missing and there may be some bugs.
Many things have been redesigned, including the UI. The Website too: NyVox.net
If you would like to join the Discord, you can do there here: https:/...
2022-06-30 12:35:33 +0000 UTC View PostA new model is being developed that will hopefully provide several improvements. I am looking for some feedback and if I should prioritize some things. A speaker's identity is a rather complex thing and separating aspects of it is a non-trivial matter.
Given the option of high speaker similarity, i.e. little information about your own voice is kept and you sound more like the target voice, and the opposite, you sound less like the target but more information is kept, things like accent ...
2022-01-24 18:25:45 +0000 UTC View Post