presamp.ini Specifications ver 1.7

11 Jul 2021 | Source | Translator: KLAD

Tags: , , ,


Advisory

When this doc was written, the presamp preset editor hadn’t been released yet. This can still help people who want to write a file from scratch, edit existing files, or understand the preset editor in more detail.
We can’t guarantee your files will work perfectly.

If you delete the default presamp.ini file, a new one will be generated using the specified default values.
Blank lines will accidentally be loaded in. (might be fixed in ver1.7)

Preset version compatibility

ver1.2:
presamp ver 0.64 and up
predit ver 1.20 and up

ver1.3:
presamp ver 0.830 and up
predit ver 1.30 and up

ver1.4:
presamp ver 0.841 and up
predit ver 1.40 and up

ver 1.5:
presamp ver 0.850 and up
predit ver 1.50 and up

ver 1.5 (features for English voicebanks, only in per-voicebank presets):
presamp ver 0.865 and up

ver 1.6:
presamp ver 0.866 and up
predit ver 1.6 and up

ver 1.7:
presamp ver 0.890 and up
predit ver 1.7 and up

File location

Until presamp ver 0.714
%appdata%\UTAU\presamp\presamp.ini

Until presamp ver 0.801
%appdata%\UTAU\presamp_DV\presamp.ini

Until presamp ver 0.845
%appdata%\UTAU\presamp_DV08\presamp.ini

Until presamp ver 0.871
%appdata%\UTAU\presamp_DV0850\presamp.ini

Current version
%appdata%\UTAU\presamp_DV0890\presamp.ini

You can also create a presamp.ini file in a voicebank folder (same location as readme.txt etc.) that contains settings for individual sections.

Format

[ENTRY]
parameter
parameter
parameter
parameter
[ENTRY]
parameter
...

[VERSION]

Default value: 1.7

Version information for the presamp.ini file.
Current valid parameters are 1.2 or 1.3 or 1.4 or 1.5 or 1.6 or 1.7 only.
presamp 0.830DV and up is 1.3
presamp 0.841DV and up is 1.4
presamp 0.850DV and up is 1.5
presamp 0.866DV and up is 1.6
presamp 0.890DV and up is 1.7

[LOCALE]

Default value: 0

Display langauge for predit plugin.
0 - Japanese
1 - English
2 - Korean
3 - Chinese (Traditional)

Error messages are only available in Japanese.
Support for other languages by reading from an external file is planned for the future.

[RESAMP]

Default values:
C:\Program Files\UTAU\resampler.exe
C:\Program Files\UTAU\resampler.exe

Full path to a resampler.
Row 1 is the default resampler
Row 2 is used with the /1 flag
Row 3 is used with the /2 flag
… and so on.
It hasn’t been fully tested, but theoretically you can have infinite resamplers.

If the exe file doesn’t exist, presamp will search for a file named resampler.exe in the same folder as UTAU.
If a flag is used with a number higher than the total number of resamplers, the default resampler will be used.

[TOOL]

Default value: C:\Program Files\UTAU\wavtoolex.exe

Full path to a wavtool.

If the exe file doesn’t exist, presamp will search for a file named wavtool.exe in the same folder as UTAU.

[VOWEL]

Default values:

a=あ=ぁ,あ,か,が,さ,ざ,た,だ,な,は,ば,ぱ,ま,ゃ,や,ら,わ,ァ,ア,カ,ガ,サ,ザ,タ,ダ,ナ,ハ,バ,パ,マ,ャ,ヤ,ラ,ワ=100  
e=え=ぇ,え,け,げ,せ,ぜ,て,で,ね,へ,べ,ぺ,め,れ,ゑ,ェ,エ,ケ,ゲ,セ,ゼ,テ,デ,ネ,ヘ,ベ,ペ,メ,レ,ヱ=100  
i=い=ぃ,い,き,ぎ,し,じ,ち,ぢ,に,ひ,び,ぴ,み,り,ゐ,ィ,イ,キ,ギ,シ,ジ,チ,ヂ,ニ,ヒ,ビ,ピ,ミ,リ,ヰ=100  
o=お=ぉ,お,こ,ご,そ,ぞ,と,ど,の,ほ,ぼ,ぽ,も,ょ,よ,ろ,を,ォ,オ,コ,ゴ,ソ,ゾ,ト,ド,ノ,ホ,ボ,ポ,モ,ョ,ヨ,ロ,ヲ=100  
n=ん=ん=100  
u=う=ぅ,う,く,ぐ,す,ず,つ,づ,ぬ,ふ,ぶ,ぷ,む,ゅ,ゆ,る,ゥ,ウ,ク,グ,ス,ズ,ツ,ヅ,ヌ,フ,ブ,プ,ム,ュ,ユ,ル,ヴ=100  
N=ン=ン=100

Vowel settings.
This section can be configured per-voicebank.

Format:
v=V=CV,CV,CV,...=vol

When the end of the previous note contains the specified CV, the following VC or VCV note will use the v.
The specified CVs will match the ending of the note lyric. For example, pa matches pa and spa but not pha.

The V setting is for sustained vowels and end breaths, but this feature is not yet usable. The planned format would use the for notes aliased like a あ息R or あー.

The vol setting can be used to set the volume for each vowel.

[CONSONANT]

Default values:

ch=ch,ち,ちぇ,ちゃ,ちゅ,ちょ=1  
gy=gy,ぎ,ぎぇ,ぎゃ,ぎゅ,ぎょ=1  
ts=ts,つ,つぁ,つぃ,つぇ,つぉ=1  
ty=ty,てぃ,てぇ,てゃ,てゅ,てょ=1  
py=py,ぴ,ぴぇ,ぴゃ,ぴゅ,ぴょ=1  
ry=ry,り,りぇ,りゃ,りゅ,りょ=1  
ny=ny,に,にぇ,にゃ,にゅ,にょ=0  
r=r,ら,る,れ,ろ=1  
hy=hy,ひ,ひぇ,ひゃ,ひゅ,ひょ=0  
dy=dy,でぃ,でぇ,でゃ,でゅ,でょ=1  
by=by,び,びぇ,びゃ,びゅ,びょ=1  
b=b,ば,ぶ,べ,ぼ=1  
d=d,だ,で,ど,どぅ=1  
g=g,が,ぐ,げ,ご=1  
f=f,ふ,ふぁ,ふぃ,ふぇ,ふぉ=0  
h=h,は,へ,ほ=0  
k=k,か,く,け,こ=1  
j=j,じ,じぇ,じゃ,じゅ,じょ=0  
m=m,ま,む,め,も=0  
n=n,な,ぬ,ね,の=0  
p=p,ぱ,ぷ,ぺ,ぽ=1  
s=s,さ,す,すぃ,せ,そ=0  
sh=sh,し,しぇ,しゃ,しゅ,しょ=0  
t=t,た,て,と,とぅ=1  
w=w,うぃ,うぅ,うぇ,うぉ,わ,を=0  
v=v,ヴ,ヴぁ,ヴぃ,ヴぅ,ヴぇ,ヴぉ=0  
y=y,いぃ,いぇ,や,ゆ,よ,ゐ,ゑ=0  
ky=ky,き,きぇ,きゃ,きゅ,きょ=1  
z=z,ざ,ず,ずぃ,ぜ,ぞ=0  
my=my,み,みぇ,みゃ,みゅ,みょ=0

Consonant settings.
This section can be configured per-voicebank.

Format:
c=CV,CV,...=[crossfade flag](=[VCLENGTH])

When inserting VC notes, presamp uses the c corresponding to the CV of the following note.
Unlike the vowel settings, the CV should match the whole note lyric exactly.

The value of [crossfade flag] should be 0 or 1.
When set to 1, the VC and the CV will not be crossfaded together.
You can use this for plosive consonants where only silence would be crossfaded, or other consonants where a crossfade would sound strange.

ver1.4 and up
[VCLENGTH] is an optional parameter. When omitted, the value is 0.
0 - Use the length of the CV’s preutterance to set the length of the VC note
1 - Use the length of the VC’s entire OTO to set the length of the VC note

[PRIORITY]

Default values: k,ky,g,gy,t,ty,d,dy,ch,ts,b,by,p,py,r,ry

This setting will prioritize using CVVC when both VCV and CVVC are present.
This section can be configured per-voicebank.

Format:
CV,CV,CV,...

Without setting priority, the notes [あ][k] would be converted to [あ][a k] and the “k” would be inaudible. Setting priority would convert the notes to [あ][a k][k] instead.

The CV matches the whole note lyric. Since VCV is ignored entirely, if CVVC samples cannot be found in the voicebank, the notes will remain CV.

In per-voicebank settings, you may want to circumvent some VCV samples by automatically forcing CVVC, but this will happen regardless of the initial vowel.
ex. To avoid using [o も], is added to PRIORITY. A note that would normally be converted to [a も] would also be forced to become CVVC, but if the alias [a m] can’t be found in the voicebank’s OTO, it would end up becoming CV.

[REPLACE]

Default values: Omitted due to length
(Converts romaji to hiragana, and converts を to お)

This setting will replace one lyric with another.
This section can be configured per-voicebank.

Format: CV1=CV2

CV1 will be replaced with CV2.

This could be useful for romaji USTs, the voicebank being used by non-Japanese users, replacing phonemes that are accidentally missing in the voicebank, or joke voicebanks like Homokure.

[ALIAS]

Default values:

VCV=%v%%VCVPAD%%CV%
BEGINING_CV=-%VCVPAD%%CV%
CROSS_CV=*%VCVPAD%%CV%
VC=%v%%vcpad%%c%,%c%%vcpad%%c%
CV=%CV%,%c%%V%
C=%c%
LONG_V=%V%ー
VCPAD= 
VCVPAD= 
ENDING1=%v%%VCPAD%R
ENDING2=-

These settings have been described since preset ver1.5, but are only compatible with preset ver 1.7.

This section sets the format for each alias type.
%v% and %V% reference the VOWEL section.
%c% and %CV% reference the CONSONANT section.
If [VCPAD], [ENDTYPE1], or [ENDTYPE2] are set in their individual sections, those settings are preferred over the ones in this section.

This section can be configured per-voicebank.

[PRE]

Default value: (none)

Format:

prefix1
prefix2
...

This functions in preset ver1.5 and up.

This setting helps presamp correctly distinguish prefixes and lyrics in notes like [A3あ].
Each prefix should be written on a separate line.

This section can be configured per-voicebank.

[SU]

Default value: %num%%append%%pitch%

Duplicate numbers: %num% (the 2 in [あ2])
Append names: %append% (the 強 in [あ強])
Pitch names: %pitch% (the G4 in [あG4])
This setting helps presamp understand the order of suffixes.

This functions in preset ver1.5 and up.

At this point in time, the default order for alias output is fixed.
Please wait for future functionality.

This section can be configured per-voicebank.

[NUM] [APPEND] [PITCH]

This functions in preset ver 1.5 and up.

A backup list of every number, append name, and pitch name.

Special commands

@UNDERBAR@
When this is set, presamp will recognize an underscore before the specified suffixes.

Without @UNDERBAR@
[あ_G4] -> Only “G4” is recognized as a suffix, and the note lyric is “あ_”
With @UNDERBAR@
[あ_G4] -> “_G4” is recognized as a suffix, and the note lyric is “あ”

@NOREPEAT@
When this is set, presamp will not recognize repeated suffixes.

Without @NOREPEAT@
[あ22] -> The suffix is “22”
With @NOREPEAT@
[あ22] -> The suffix is “2”

This section can be configured per-voicebank.

[ALIAS_PRIORITY] [ALIAS_PRIORITY_DIFAPPEND] [ALIAS_PRIORITY_DIFPITCH]

This functions in preset ver 1.7 and up.

This sets the priority of alias format for normal cases, for consecutive notes with different appends, and consecutive notes using different recorded pitches.

Default values:

VCV
CVVC
CROSS_CV
CV
BEGINING_CV

However, VCV and CVVC are switched for different appends and different pitches.
The format of each alias type is specified in the ALIAS section.

This section can be configured per-voicebank.

[SPLIT]

Default value: 1

This section can be configured per-voicebank.
This functions in preset ver 1.5 and up.

When a note contains multiple consonants and vowels, presamp can automatically split them into more notes.
Setting this to 0 will disable note splitting.

[MUSTVC]

Default value: 0

This section can be configured per-voicebank.
This functions in preset ver 1.6 and up.

When converting to CVVC, if the vowel of the CV is less than 20 msec, a VC will not be played. (presamp specification)
If this setting is set to 1, a VC will be inserted no matter how short the CV is.
The length of the VC will be half of the CV, rather than the length of the following note’s preutterance.

Set this to 1 for voicebanks with diphthongs, such as English or Chinese banks.

[CFLAGS]

Default value: p0

This flag will automatically be applied to consonant-only notes, or consonant cluster notes. This section can be configured per-voicebank.

This functions in preset ver 1.3 and up.
If this is specified in preset ver 1.2, presamp will crash.

The consonants that are recognized are specified in the CONSONANT section.

[VCLENGTH]

Default value: 0

This functions in preset ver 1.4 and up.
If this is specified in preset ver 1.2, presamp will crash.

This setting determines the length of VC notes when converting to CVVC.
0 - Length of preutterance in the following CV
1 - Length of the whole VC OTO

If this is set to 1, any vclength settings in the CONSONANT section will be ignored.

[ENDTYPE1] [ENDTYPE2]

Default value: (none)

Format for ending notes. These are only used when ENDFLAG is set to 3.
You can only set this in preset version 1.5 for English voicebanks.

When converting rests, ENDTYPE1 is used.
When adding text to the end of an alias, ENDTYPE2 is used.
The format is the same as ENDTYPE.

In ver 1.7 and up this has been integrated into ALIAS, but this remains for backwards compatibility.

[ENDTYPE]

Default value: %v% R

Format for ending notes.
This section can be configured per-voicebank.

This functions in preset ver 1.3 and up.
The behavior depends on the value of the ENDFLAG parameter described below.

In ver 1.7 and up this has been integrated into ALIAS, but this remains for backwards compatibility.

%v%: The “a” “i” “u” etc. in the VOWEL section
%V%: The “あ” “い” “う” etc. in the VOWEL section
There is planned support for other characters used as-is.

[VCPAD]

Default value: ` `
(space)

This section can be configured per-voicebank.

This feature is for an old request that went “Since CV is written as ka, I want VC formatted as ak.”
The format of VCs is %v%%vcpad%%c%.

This is mainly for non-Japanese users.
It’s suddenly possible to have VC aliases like [a+k], but this hasn’t been fully tested.

If this section is left blank, presamp will crash, so it’s not usable.

In ver 1.7 and up this has been integrated into ALIAS, but this remains for backwards compatibility.

[ENDFLAG]

Default value: 1

The default value in ver 1.3 and up is 0

This functions in preset ver 1.3 and up.
This setting specifies whether or not to automatically use ending notes.
This section can be configured per-voicebank.

0 will disable automatic ending notes

1 will convert rest notes to end notes
ex. When ENDTYPE is %v% R, the notes [あ][R] will be converted to [あ][a R]

2 will convert the last note to an end note
ex. When ENDTYPE is -, the notes [a k][R] will be converted to [a k-][R]

This setting is only supported by preset ver 1.5 for English voicebanks and up.
3 will do the function of both 1 and 2.

[BATNUM]

Default value: 16

Presamp’s acceleration setting.
The number ranges from 1 to 2x the number of CPU cores.

No matter how large you set this, the actual number will always be limited to 2x the number of CPU cores.