How to Use & Set-Up Presamp/AutoCVVC 2.0 for UTAU

Makku & Violin

Makku's SoundCloud Makku's Twitter
Violin's SoundCloud Violin's Twitter

Hello! So, just a quick disclaimer, this guide was written using both Makku's and Violin's banks as a base and there's no guarantee that this will work perfectly with every single bank. For the most part this should work with most banks, baring weird aliasing systems or errors. 

If you would like a video version of this tutorial, please see Here

Downloads

Recommended Downloads

• presamp 0.8996DV (Recommended option)
  https://ch.nicovideo.jp/delta_kimigatame/blomaga/ar477661
• autoCVVC 2.002
  https://ch.nicovideo.jp/delta_kimigatame/blomaga/ar1172141
  

Both of these tools, along with the base for the reclist for the Japanese voicebank were made by Delta, go check out their blog, they make tons of cool material for UTAU!

Extra features (check UST in extra folder): (For Reese JPN)

• Glottal stops (triggered by ・ or . in romaji);
  
• Vocal fries (triggered by ');
  
• Breaths (吸, 吸vowel, brvowel, brnumber, br, breath);
  
• Consonant sounds (k, ky, g, gy, t, ty, d, dy, ch, ts, b, by, p, py, r, ry, sh, s);
  
• L and V addons (ヴぁ, ラ, etc...)
  
• Falsetto (triggered by F);
  
• Romaji.

Extra features (check UST in extra folder): (For Reese ENG)

• Glottal stops (triggered by .);
  
• Vocal fries (triggered by ');
  
• Presamp/autoCVVC compatibility;
  
• Endbreaths (triggered by h).

How to use Presamp:

​Download presamp from the link provided above. Open "presamp08996.zip" and put the files "wavtoolex.exe" and "presamp.exe" in UTAU's installation folder. Open the folder "hook4presamp20140614" and extract the "dummy" folder inside it in UTAU's plugin folder (you can find it by opening UTAU, going to "Tools(T)", then "Plug-Ins(N)" and then clicking on "Open Plug-Ins Folder(O)"). Open "predit1730.zip" and extract the "predit" folder in the same plugin folder.

​IMPORTANT: If you haven't already, please make sure your Locale is set to Japanese, that your time format is also set to Japanese and that the decimal symbol is a . (dot) as opposed to , (comma). Apparently only the first step is necessary for American systems, but if you're having issues try all of them.

​Go to "Tools(T)", then "Option(O)..." and change these options:
- Rendering| Turn off all options except for the first one (first one is optional);
- Cache| Turn on "Cache intermediate files". (Remove cache files at quit is recommended)

​That should be it! Open your UST (both CV and VCV will work, but VCV sounds less smooth on my end, so I recommend you convert the UST to CV format, doesn't matter if romaji or hiragana), choose the voicebank, set wavtool and resampler BOTH as "presamp.exe" and make sure to go to the Plugins and open the dummy plugin! If presamp is glitching out you can use the dummy plugin each time to fix it. You can use the predit plugin to change which wavtool and resampler you'd like to use. Additionally, if presamp keeps crashing try to go into the predit plugin and setting the number of bats to a lower number.

Additional notes for the English banks:

• The voicebank (Reese ENG, found here on Reese's page) has full support for presamp! To use it simply type the phonetics you wish (list below) for each syllable.
  e.g.: This is Reese = [DIs] [Iz] [ris]
  
• Ending consonants/vowels are made automatically by presamp or autoCVVC so no need to worry about those.
  
• If you don't wanna use presamp, we recommend you type in the phonetics, then use autoCVVC on each sentence separately to fine tune as needed. It's also important to go slow when using autoCVVC as it has a tendency to mess up timing. 
  
• IMPORANT NOTE: [l] is only applicable to words/syllables that start with l, for everything else PLEASE use [5]

Known issues:

• Consonant clusters are hit or miss, I suggest adding a separate note for them (issue only present in LITE reclist).
  Recommended solution: cry = [kr] [aI] or [kr] [raI]. [kraI] may work also work, but it often glitches out.
  
• Vowel extensions aren't functional for some reason we haven't figured out. If you wanna extend a vowel you may encounter some issues.
  Recommended solution: using !(vowel). The [!] makes it so that presamp stops being "automatic" for that specific note.
  
• Number errors are caused when a specific phonetic combo can also resemble a voicebank's pitch.
  Recommended solution: try avoiding using combos like E4 or A5, for example by putting numbers in the next note.
  e.g.: better = [bE4] [3] this is likely to glitch, so try [bE] [43]
  
• Lack of P0. Presamp has a feature that allows the flag "P0" to be used for specific notes (mostly clusters) to avoid volume contrast. However through our testing this feature wasn't working. This isn't a major issue, but it's worth noting.
  Recommended solution: ignore the issue or use P0 globally (this may make mixing harder, but will probably result in better quality overall)
  IMPORTANT: not all resamplers support the P0 flag. P0 cancels UTAU's normalization, so that the volume is equal to the original recording's. For example, tn_fnds does not support this flag.
  
• Instability. This method is mostly working fine, but if it may be unstable on certain notes, especially in regards to consonants.
  Recommended solution: split the consonants manually. This may be especially needed on faster songs.
  e.g.: picture = [pI] [k] [tS] [3]

How to use AutoCVVC 2.0:

​Download autoCVVC from the link provided above. Open "autoCVVC2.002.zip" and extract the "autoCVVC2.002" folder in UTAU's plugin folder (refer to presamp section for where to find it). Open UTAU and select the notes you'd like to convert to CVVC, then open the autoCVVC plugin.

1. Change the language from "Ja" to "En" (bottom left).
2. Make sure these options are selected (these may change depending on the bank):

• Optimize;
• Use [-CV];
• Use ending note;
• Allow replace;
• Allow split;
• Use presamp; (optional, better if used with combination with Presamp... if that's your sort of thing...)
• Adjust param (optional, but recommended. Though this can be quite buggy and if you experience glitches it is recommended that you clear the UST and not use this option/the option below for the remainder of the UST);
• Crossfade (optional, but recommended. Though this can be quite buggy and if you experience glitches it is recommended that you clear the UST and not use this option/the option above for the remainder of the UST);.

Click submit and you're done!

Additional notes for the English banks:

It is recommended that you use Presamp for the English VBs but if you want to use AutoCVVC, we recommend going very slow and checking constantly for errors. Also, it should be noted that we have not done extensive testing with AutoCVVC as our main goal was Presamp support. It may preform differently or similarly to Presamp but thus far our testing has shown us that it is very similar.