Introduction
When I was first learning how to use UTAU, I wanted to use an English voicebank. In the process of looking for a good one, I saw a few recommendations for Adrian. After listening to his demo songs on YouTube by their creator Adriann_V, I fell in love with the androgynous and choral sound of his voice. There was only one problem: Adrian_Medium uses a proprietary VCV English format, meaning there were almost no guides for it anywhere. There was a video by Adriann_V explaining things a little, but it was confusing for me.
That video has since been taken down, and while I do have a copy recorded in 360p resolution, I will not be reuploading it due to it containing what I believe is Adriann_V’s actual voice. Instead, I decided to take what the video said and combine it with what I have learned from using him and make my own guide for those who wish to learn how Adrian_Medium works.
DISCLAIMER: None of this is endorsed by Adriann_V themselves, and on the off-chance they see this and wish for me to take it down, I will. As well, this does not apply to Adrian_Possessed, Exorcism, or the unfinished Repentance. I do not know how those work, as they are different.
And with that out of the way, let’s get started!
Section 1: Vowels
Alright, let’s start with vowels. Adrian_Medium has 11 vowels, and they all serve different purposes for the 4 languages Adrian_Medium is compatible with. For now, I will try to compare them as best I can to the Vocaloid X-Sampa phonemes, as I am more familiar with them compared to Arpabet.
-----------------------------------------------------------------------------------------------------------------------------------
[@] = [{] (Pronounced Trap)
[a] doesn’t directly correspond to a Vocaloid phoneme, but is closest to the A in [aI]
[aw] = [O:] (Pronounced like Taught)
[e] = [e] (Pronounced like Them)
[er] is strange, sort of like [@r] but with less “r” coloring
[eu] is the Japanese phoneme [M], their “u” phoneme
{i} = [i:] (Pronounced like Beef)
[ih] = (Pronounced like Kit)
[o] is similar to [a] in that it doesn’t directly correspond to a Vocaloid phoneme, but it is closest to the beginning of [OI]
{u} = [u:] (Pronounced like Boot)
[uh] = [V] or [@] (Pronounced like Aware)
-----------------------------------------------------------------------------------------------------------------------------------
As you may have noticed, I did not list any diphthongs here. That’s because Adrian_Medium doesn’t have any! That’s right, Adrian does not have separate samples for diphthongs and needs to use VV vowel transitions to make these sounds. Fortunately, these are pretty easy to remember once you get used to
them. I’ll list a few of the most common ones here.
-----------------------------------------------------------------------------------------------------------------------------------
[eI] or the word “A” = [e] [e i]
[aI] or the word “I” = [a] [a i]
[OI] or the “oy” in “Boy” = [o] [o i]
[@U] or the start of the word “Oat” = [o] [o u]
[aU] or the “ou” in the word “Loud” = [a] [a u]
-----------------------------------------------------------------------------------------------------------------------------------
As well, as far as I have encountered, Adrian_Medium does not have any “R” accented vowel phonemes. This is pretty easy to fix though. Just attach a “R” phoneme of your choosing to the end of the vowel, and you’ll be good to go. And speaking of consonants, let’s move onto Section 2!
Section 2: Consonants
Next up are the consonants, and some of them are a bit strange. As a rule of thumb, if you can sound out a word and try to spell it from there, the consonants will be right. For example, the word “behind” is spelled phonetically something like [bi] [i ha] [a i] [i nd]. There are a few notable exceptions to this that you will want to know. Because English over-pronounces a lot of consonants, Adriann_V recorded phonemes for English consonants separate from the Japanese and Spanish ones. I’ll list some of them below.
-----------------------------------------------------------------------------------------------------------------------------------
T - [tt]
L - [ll]
K - [kk]
D - [dz]
P is unique as it ONLY has a doubled version for CCV phonemes, for example [ppro] vs [pro]
-----------------------------------------------------------------------------------------------------------------------------------
In addition, Adrian_Medium has 3 versions of a phoneme for “R”: [r], which is the Japanese “R/L” combo platter; [rr], which is the Spanish rolled “R”; and [3], which is the English hard “R”. One thing recommended by Adriann_V is to use the Japanese “R” (which is [r]) instead of [tt] for smoother transitions, similar to the SynthV phoneme [dx]. I don’t think Vocaloid has a version of this, but it might be something like [d].
Section 3: Miscelleneous
And finally, this is a section devoted to random tips and tricks I picked in my time using Adrian. I hope these help someone! Fair warning, I use OpenUTAU, so I don’t know if they will work in Classic UTAU. Enjoy!
Something of note are the syllables [kk] and [tt]. At the end of a word where you would like to use one of those, be careful. If you’re looking for a slight “T” or “K” sound on the end, then just putting a vowel to consonant transition would be fine. However, if you’re looking for a hard “T” or “K” sound, you’ll have to change some things. For some odd reason, just putting a vowel to consonant transition does not actually pronounce the plosive, it more puts a slight suggestion of the plosive on the end of the vowel. I’ve found that turning the VC transition into a CC phoneme really emphasizes them for some reason, for example the word “take” If I wanted a softer “K”, I’d put [tte] [e i] [i kk]. If I wanted a hard “K” I’d put [tte] [e i] [-kk]. The only problem with this is that it can sound kind of choppy with the next note, and the vowel sound before it – in this case the sound – fades out long before the [kk] comes in. To remedy the first issue, I have no idea sorry. For the second, in OpenUTAU there’s an option to view the waveform of the notes. Do so and make note of where the wave trails off before the [kk]. If you add in this case [i i] by splitting the note, it’ll extend the sound. I hope that made sense. Remember, this also works for [tt].
When trying to make diphthongs, be careful. Ideally, you want the second half of the diphthong to be about half the length of the first half. For example, the diphthong [aI] would be [a] [a i]. The [a] alone should take up about 2/3 or 3/4 of the note, then you put the [a i] at the remaining third or fourth. DO NOT, I repeat DO NOT MAKE THEM EQUAL LENGTHS. That is NOT how diphthongs work, your singing will sound unnatural and like improper English. This applies for all diphthongs, not just [aI].
When using a Japanese VCV UST file, make sure to remove any notes labeled with either a dash in front of them or a transition from the “n” phoneme. This is because Adrian does not have any phonemes labeled with the dash nor any “n” transitions. Do I know why? No idea. For example, the phoneme [- mi] becomes [mi], and [n ka] becomes just [ka]. It might sound strange, but you’ll have to work around it.
I’d also recommend switching any instance of a Japanese phoneme with “u” in it with [eu]. This is because is an English phoneme, and [eu] is the Japanese “u”. Of course, this is not necessary if you want Adrian to have an accent of some sort, but if you want to minimize that sort of thing try switching it.
I think that’s all the tips I have, have fun using Adrian! I hope you like their voice and choose to stick with them!
When I was first learning how to use UTAU, I wanted to use an English voicebank. In the process of looking for a good one, I saw a few recommendations for Adrian. After listening to his demo songs on YouTube by their creator Adriann_V, I fell in love with the androgynous and choral sound of his voice. There was only one problem: Adrian_Medium uses a proprietary VCV English format, meaning there were almost no guides for it anywhere. There was a video by Adriann_V explaining things a little, but it was confusing for me.
That video has since been taken down, and while I do have a copy recorded in 360p resolution, I will not be reuploading it due to it containing what I believe is Adriann_V’s actual voice. Instead, I decided to take what the video said and combine it with what I have learned from using him and make my own guide for those who wish to learn how Adrian_Medium works.
DISCLAIMER: None of this is endorsed by Adriann_V themselves, and on the off-chance they see this and wish for me to take it down, I will. As well, this does not apply to Adrian_Possessed, Exorcism, or the unfinished Repentance. I do not know how those work, as they are different.
And with that out of the way, let’s get started!
Section 1: Vowels
Alright, let’s start with vowels. Adrian_Medium has 11 vowels, and they all serve different purposes for the 4 languages Adrian_Medium is compatible with. For now, I will try to compare them as best I can to the Vocaloid X-Sampa phonemes, as I am more familiar with them compared to Arpabet.
-----------------------------------------------------------------------------------------------------------------------------------
[@] = [{] (Pronounced Trap)
[a] doesn’t directly correspond to a Vocaloid phoneme, but is closest to the A in [aI]
[aw] = [O:] (Pronounced like Taught)
[e] = [e] (Pronounced like Them)
[er] is strange, sort of like [@r] but with less “r” coloring
[eu] is the Japanese phoneme [M], their “u” phoneme
{i} = [i:] (Pronounced like Beef)
[ih] = (Pronounced like Kit)
[o] is similar to [a] in that it doesn’t directly correspond to a Vocaloid phoneme, but it is closest to the beginning of [OI]
{u} = [u:] (Pronounced like Boot)
[uh] = [V] or [@] (Pronounced like Aware)
-----------------------------------------------------------------------------------------------------------------------------------
As you may have noticed, I did not list any diphthongs here. That’s because Adrian_Medium doesn’t have any! That’s right, Adrian does not have separate samples for diphthongs and needs to use VV vowel transitions to make these sounds. Fortunately, these are pretty easy to remember once you get used to
them. I’ll list a few of the most common ones here.
-----------------------------------------------------------------------------------------------------------------------------------
[eI] or the word “A” = [e] [e i]
[aI] or the word “I” = [a] [a i]
[OI] or the “oy” in “Boy” = [o] [o i]
[@U] or the start of the word “Oat” = [o] [o u]
[aU] or the “ou” in the word “Loud” = [a] [a u]
-----------------------------------------------------------------------------------------------------------------------------------
As well, as far as I have encountered, Adrian_Medium does not have any “R” accented vowel phonemes. This is pretty easy to fix though. Just attach a “R” phoneme of your choosing to the end of the vowel, and you’ll be good to go. And speaking of consonants, let’s move onto Section 2!
Section 2: Consonants
Next up are the consonants, and some of them are a bit strange. As a rule of thumb, if you can sound out a word and try to spell it from there, the consonants will be right. For example, the word “behind” is spelled phonetically something like [bi] [i ha] [a i] [i nd]. There are a few notable exceptions to this that you will want to know. Because English over-pronounces a lot of consonants, Adriann_V recorded phonemes for English consonants separate from the Japanese and Spanish ones. I’ll list some of them below.
-----------------------------------------------------------------------------------------------------------------------------------
T - [tt]
L - [ll]
K - [kk]
D - [dz]
P is unique as it ONLY has a doubled version for CCV phonemes, for example [ppro] vs [pro]
-----------------------------------------------------------------------------------------------------------------------------------
In addition, Adrian_Medium has 3 versions of a phoneme for “R”: [r], which is the Japanese “R/L” combo platter; [rr], which is the Spanish rolled “R”; and [3], which is the English hard “R”. One thing recommended by Adriann_V is to use the Japanese “R” (which is [r]) instead of [tt] for smoother transitions, similar to the SynthV phoneme [dx]. I don’t think Vocaloid has a version of this, but it might be something like [d].
Section 3: Miscelleneous
And finally, this is a section devoted to random tips and tricks I picked in my time using Adrian. I hope these help someone! Fair warning, I use OpenUTAU, so I don’t know if they will work in Classic UTAU. Enjoy!
Something of note are the syllables [kk] and [tt]. At the end of a word where you would like to use one of those, be careful. If you’re looking for a slight “T” or “K” sound on the end, then just putting a vowel to consonant transition would be fine. However, if you’re looking for a hard “T” or “K” sound, you’ll have to change some things. For some odd reason, just putting a vowel to consonant transition does not actually pronounce the plosive, it more puts a slight suggestion of the plosive on the end of the vowel. I’ve found that turning the VC transition into a CC phoneme really emphasizes them for some reason, for example the word “take” If I wanted a softer “K”, I’d put [tte] [e i] [i kk]. If I wanted a hard “K” I’d put [tte] [e i] [-kk]. The only problem with this is that it can sound kind of choppy with the next note, and the vowel sound before it – in this case the sound – fades out long before the [kk] comes in. To remedy the first issue, I have no idea sorry. For the second, in OpenUTAU there’s an option to view the waveform of the notes. Do so and make note of where the wave trails off before the [kk]. If you add in this case [i i] by splitting the note, it’ll extend the sound. I hope that made sense. Remember, this also works for [tt].
When trying to make diphthongs, be careful. Ideally, you want the second half of the diphthong to be about half the length of the first half. For example, the diphthong [aI] would be [a] [a i]. The [a] alone should take up about 2/3 or 3/4 of the note, then you put the [a i] at the remaining third or fourth. DO NOT, I repeat DO NOT MAKE THEM EQUAL LENGTHS. That is NOT how diphthongs work, your singing will sound unnatural and like improper English. This applies for all diphthongs, not just [aI].
When using a Japanese VCV UST file, make sure to remove any notes labeled with either a dash in front of them or a transition from the “n” phoneme. This is because Adrian does not have any phonemes labeled with the dash nor any “n” transitions. Do I know why? No idea. For example, the phoneme [- mi] becomes [mi], and [n ka] becomes just [ka]. It might sound strange, but you’ll have to work around it.
I’d also recommend switching any instance of a Japanese phoneme with “u” in it with [eu]. This is because is an English phoneme, and [eu] is the Japanese “u”. Of course, this is not necessary if you want Adrian to have an accent of some sort, but if you want to minimize that sort of thing try switching it.
I think that’s all the tips I have, have fun using Adrian! I hope you like their voice and choose to stick with them!