(for better understanding and information regarding

recording and otoing I recommend

reading the Google Document​ above,

if you have questions, I will try my best to help you!)

​

• languages: English, German, French, Chinese, Japanese, Spanish

• recordings: 181 (each pitch), about 6 syllables (each line)​

​​

Making Usts:

​

For Japanese Usts, you can simply use Autocvvc with the default preset and it will automatically turn every Hiragana-CV-Ust into a CVVC-Ust that is singable for your voicebank. For any other language lyric input is very similar to other English VCCV Ust, except that the aliases and the lyrics input have visually more similarities to Arpasing Usts. For reference and easier understanding I included an Example Ust in the folder with the Reclist! Also, here is a random sentence and how you would write out the syllables for your voicebank: “I spend”

​

“- ah” “ah y” “y s” “s p” “b eh” “eh n” “n d0”

​

Here are the three things that you need to pay attention to:

• The diphthongs ay, ey, oy, aw, ow are divided into two phonemes in order to shorten the reclist. For example:

  • "ay" => "ah y" This means, if you want your voicebank to sing: "I fell down." you would write

  • "- ah" "ah y" "y f" "f eh" "eh l" "l d" "d ah" "ah w" "w n" "n -"​

  • "y f" is the same file as "ih f", and the "ah y" is the same file as "ah ih", just with a different alias.

  • This doesn't seem to worsen the pronunciation that much and the transitions are barely noticeable, I think.

• The reclist has double consonant endings, which are currently marked with a zero: "m d0" or “R t0” for example. These are used if a rest follows the double consonant, similar to “md-” and “rt-” in VCCV. (I don’t know why I chose to use a zero instead) For single consonants, however, you would write “t -” or “ch -”.

• Right now, you would need to write "s p" "b ih" "ih n" for words like “spin”, since the “p” would sound similar to a “b” in the pronunciation of "spin", but it's possible to add an extra consonant for that. This also applies to “s k” “g ah” “ah y” for the word “sky”.

​​

The phonemes are listed on the right side. You can decide to not record the lines intended for the extra phonemes, if you won't need them or add new consonants, for example if you want to differentiate between aspirated or unaspirated consonants.

​

Here's a demo using my own Multilingual voicebank:

​