Can't build Voice Models/Bank?

I keep getting these errors whenever I try to build my voice models and voicebank. These screenshots show some of the things I’ve tried to do to fix it, but none of them have worked. I’ve also tried replacing the romaji symbols in the model symbols box with hiragana and I’ve even tried renaming the hiragana wav files to romaji and nothings working.
Also, I don’t know if this is relevant, but whenever I close the voice configure window it deletes all of the configuring work I did even if I save the whole project. It’s not a huge problem since it’s pretty easy, but it’s a bit frustrating since it does take a while because I have every sound saved to it’s own wav.

If anyone knows how to fix it or has any ideas that might not even work, please let me know.
Or if anyone would be willing to download my voicebank and the dvtb file (that will only save the phonetic dictionary) and try to fix it, please let me know.

  • Model files does not exist:
    The model files has not been generated and cannot be found by the program. Please generate them and put the correct directory.
  • Random start and end with absolutely nothing
    Please check your Wav Location and ensure that DeepVocal Toolbox is run as an administrator.
  • Error that nothing exist in “insert location”
    I do not know how to fix this as I do not use that.
  • Pitch symbol error
    The symbol you have written is missing or wrong. Please write the correct pitch symbol. It is not written in the photo under “Build Voice Bank”.

If everything still fails, please send your voicebank to me and I will look into it.

1 Like

Thank you for responding and I’m so sorry for the late reply >.<

Ok, so I went back and double checked some things and It turned out that I had the wrong folder selected (even though I could’ve sworn that it was the right one) and I also went through and fixed some of the otoing and it now builds the voice model files without much problems other than I get an error on the za ze zo and zu’s that says “Unvoiced part exists in voiced consonant.” even though It doesn’t sound like there’s any unvoiced parts. And then when I go to build the voicebank it said that the model files don’t exist even though they do and I made sure that the right folder was selected and it is. However, when I tested the voicebank in deepvocal, it sounded mostly fine aside from the fact that a few of the sounds are completely silent.
Those sounds include all of the vowels as well as: N, za, ze, zo, and zu

Other than that there doesn’t seem to be any other problems aside from simple ones like some of the sounds being too loud or the quality of some sounds is worse than others, and a couple of sounds cutting off too early, but I’m not as concerned with those issues yet as I can probably fix those easily.

Please check your pheonic dictionary symbols to see if “za ze zo zu” exists alongside its voicebank configuration and/or generated model if any. Additionally, try putting ‘z’ inside unvoiced consonants instead. It may help.
Please check if “N” exist in your vowel list as well if you intend “N” to be used as a vowel instead of “n”, voicebank configuration and/or generated model if any.