Try creating an English song with VOCALOID5
There are mixed opinions about using VOCALOID5, and we’re still in a phase of feeling our way around its capabilities.
VOCALOID5 adds features such as the ability to select singing styles in addition to differences in parameter behavior compared to 4 and adjustments to the singing voice, greatly increasing what you can do even in a default state.
For those who refined their expression in the 4 series, some seem puzzled by the editor’s quirks and the changes in how parameters affect the sound.
Personally, with VOCALOID5 making it easier to edit areas that were hard to vary using the 4-series parameter approach, I feel that editing in VOCALOID5 brings me closer to the sound I have in mind.
Following on from the previous article, I’d like to continue exploring what kinds of songs can be sung with VOCALOID5.
English library Amy
last timeI tried making it sing using IA and Kaori.
We’ve found that expressiveness has improved even with the VOCALOID4 voice libraries, so this time I’d like to try singing in English.
VOCALOID5 comes with two standard libraries: Chris (male voice) and Amy (female voice).
This time, I’ll use the backing track I made for performance and have Amy sing a song called “Tomorrow is Mine.”
youtube.
com/watch?v=rQsNr13c9_o
This song is the theme for the game Bayonetta 2, which was developed by the Japanese studio PlatinumGames.
It’s a very stylish, piano-centered track where vocalist Keeley Bumford sings sometimes sultrily and sometimes powerfully—a song with striking shifts within a single chorus that makes it deeply engaging to listen to.
I copied this piece a little while ago to perform it, so although it’s a difficult piece, I’ll aim for the richest expression possible.
Basic parameter settings
Set the singer preset (what VOCALOID4 calls a Singer Preset) before entering the lyrics.
In VOCALOID5, the Style preset stores a complete set that includes both the basic singing parameters and the effects applied to the vocal.
First, we will set up the basic singing style.
The one on the far left is the SINGING SKILL.
This may be the fundamental part and possibly the most crucial aspect of VOCALOID5.
In SINIGN SKILL, by selecting the intended genre, it will add genre-appropriate basic expressions to the vocal data you input.
This time, while the genre leans a bit more toward jazz, I set it up with the idea that the vocal delivery should be soulful.
Toggling this SINGING SKILL changes various aspects such as the start of singing, vibrato, and movements during pitch transitions.
Even just selecting a SINGING SKILL that matches the musical style will change the expression, so the first step is to find the optimal way to express it.
You can choose a basic singing style by adjusting the intensity of the expressive quirks with SKILL and fine-tuning the amount applied with AMOUNT.
In VOCALOID4, something similar was possible via jobs, but since it was still in development, it didn’t achieve the level of expressiveness or appropriate effects found in VOCALOID5.
I think we can benefit from evolution even just in this part.
The VOICE COLOR in the center corresponds to the singer presets up to VOCALOID4, and it lets you select the color of the singing voice.
This part is the same as in VOCALOID4, and by making full use of CHARACTER (formerly GEN up to version 4) to adjust pitch, EXCITER to control vocal presence and clarity, and AIR to control whispery tones, it allows for a wider range of expression than in VOCALOID4.
In particular, AIR and EXCITER seem like they could be used effectively to recreate a distinctive vocal character if used well.
The actual vocal data has been created in the screen settings.
Since Amy’s natural vocal timbre differs from the original singer’s, forcing it to match would make it unpleasant to listen to. Therefore, I configured it to prioritize lyric intelligibility and dynamic expression.
English lyrics input
Unlike in Japanese, English lyrics can pack many words into short passages.
The basic rule is one note per word, though in some cases two words may be blended together.
For entering English lyrics, the steps are the same up to opening the input window: select the note where you want to start input, then choose “Flow Lyrics” from the right-click menu.
As you type words into the window separated by spaces, they are automatically assigned to notes, but longer words may not be split as intended, so those parts should be corrected manually.
For songs like this one, where the tempo is fairly fast and the phrases are sung rhythmically, there are parts that won’t work well if you just input the lyrics as-is.
Also, since the entered word will default to its common pronunciation, you may need to change the final sound to one different from the original pronunciation when you want to cut the sound off crisply.
In this song, the word “Down” in the accented part of the B section—“Bang Bang, Down Down”—doesn’t quite fit the rhythm if you sing it as “da-un” in Japanese, with the “u” stretching the timing. So, while the lyrics are written as “Down,” we’ve adjusted the pronunciation to sound like “done” (ダン) to match the rhythm.
What are the tuning results of VOCALOID5?
Broadly speaking, the basic adjustments are as described up to this point, but for finer details, we tweak various parameters and set them while listening to the results.
Since multiple parameters cannot be displayed in the window, I created a version for the video that shows the Exciter parameters, which are important for things like adding accents.
In practice, I’m also adjusting parameters like Dynamics and Pitch, and in some parts Growl and Breathiness.
youtube.
com/watch?v=6JP9O0ReIqM
It's a shame that there's a low-pitched part that inevitably sounds a bit rough, but I feel that in the other parts, my efforts are finally paying off.
I expect we’ll see more and more people letting us hear amazing songs from now on, and if the number of libraries dedicated to VOCALOID5 increases, I think it will be able to sing even more naturally.
Where will the first song using VOCALOID5 to reach one million views come from, and what kind of amazing track will it be?
I’m already looking forward to it.