STOVE Lounge

There was a special AI project that was in the news at one point. Some of you may still remember it. That's right, it was a project to train AI to listen only to a person's voice and draw the face of the person speaking .

A project where AI learns human voices and draws montages!

There was a time when I had a big dream of using this to create a composite image of a criminal using only their voices. You may have noticed that we haven't heard this news much lately. The first news story was in 2019 ...
The reason is simple: a person's appearance and voice are completely unrelated.
At the time when AI was being built, it is said that YouTube data was mainly used as learning data. That's why, in the case of celebrities that are easy to see on YouTube, they showed a 'similar-looking hit rate'.
The moment a person with a unique voice appeared, the AI started spitting out completely wrong results.
AI that creates appearances based on voices. However, it frequently shows inconsistencies and failed to be commercialized .
- The uniqueness of a voice is so unique that it is called a second fingerprint ! Acquired factors also play a big role.
The voice is so “unique” that it is even used in criminal investigations by the National Institute of Scientific Investigation. It is called a sound print that has the same characteristics as the fingerprint print of a finger. In some cases, it played a very important role in solving the case.
Since vocal cords are basically formed by the expression of genes, they tend to have similarities between families. There are also (somewhat unsubstantiated) examples of children of singers or actors working in similar industries. When other people hear the voices of family members, they certainly feel that there is a lot of similarity between parents and children.
The voice is greatly influenced by genetic factors as well as acquired factors . From simply maintaining your vocal cords to suffering from a serious illness, gaining too much weight, or even developing vocal nodules, all of these things can dramatically change your voice. This increases the “uniqueness” of the voice and widens the gap between appearance (genetics) and voice (acquired).
Yamada Noe is said to have developed a 'husky voice' after suffering from a high fever at the age of 5. Acquired factors play a very large role in the uniqueness of a voice.
- The genes that determine vocal cords and their appearance are not “linked.”
The redhead gene and the freckle gene are so closely related that they are called “linkage” genes, meaning they are inherited almost together. People with hemophilia have a very high chance of being color blind, which is also due to a linked gene.
If the genes that create human appearance (face) and the genes that determine vocal cords are linked in the same block, then the above “voice = face shape” hypothesis would be correct. Not surprisingly, there has been no reported evidence or tendency for the genes for the two traits to be physically adjacent or linked together in a block (linkage).
Even if we only consider genetic factors, we can see that there is no relationship between appearance and voice. Because of the nature of the acquired factors playing a large role, there is no correlation between the voice and a person's face or personality.
Red hair and freckles are more likely to be inherited by one gene at a time rather than through linkage inheritance.
- But why do we associate ‘character=voice’?
Shirobako Episode 14 - “Feeling uneasy? If there’s anyone who feels that way, just let them feel it!”
When watching animation or games, you've probably thought at least once, 'This character would have this kind of voice.'
A cold personality requires a low, cool tone, a cute appearance requires a bright, high tone. It's almost as if we naturally connect the character's personality to their voice.
This connection between character personality and voice is thanks to stereotypes and the dedicated acting of the voice actor who played the character . That is to say, it is just a conventional connection based on experience .
In fact, even if the voice is completely different, and even if the character's personality is completely different, we quickly adapt, "only feeling a momentary sense of discomfort." There are definitely people among your friends, coworkers, and acquaintances who have no connection between their appearance and their voice. These are cases where you become accustomed to it over time.
The characters are played by Miyuki Sawashiro (center), a voice actress known as a “wide range voice actress,” including male, female, loli, old man, husky, and even cute. If I were to pick a character that you may be familiar with, it would be the voice actor who also played Sinon from SAO and Raiden from Genshin Impact (Japan).
You will feel that there is no connection between the voice actor's appearance and the character. You can also see that not all of the characters above are connected . There are crazy characters, pure manga characters, boyish characters, and just plain loli-loli.
Yet, we don't feel much discomfort when we imagine the character's voice. That's because the character's voice is convincingly created through excellent acting . And more precisely, it says, “It was learned from the voice of the person who acted best.”
If a lively character was played by a manly and tough voice actor and became popular, The stereotype that “cheerful characters have gruff voice actors” is created in general.

If there was a tsundere character with a deep voice... I think that would be okay as is ㅋㅋㅋㅋㅋ

This is the result of learning that 'Tsundere voice = Cookie Ming'.
Just how important acting is can be easily understood by looking at some cases where people were criticized as “miscasting!”
Yuuki Kaji - Eren Yeager from "Attack on Titan": A gentle voice without any sense of tragedy Mamoru Miyano - Light Yagami from "DEATH NOTE": A voice with personality but no godlike majesty Hayashi Yu - Sosuke from "Skip and Loafer": A sleepy, weak voice
At first, everyone was criticized for choosing the wrong voice actors. These are voices that have now become established as unique voices that immediately come to mind when you imagine the character .
As planned ㅋㅋㅋㅋ At the time of casting, he was called a “miscasting” because his voice was too unique and didn’t have the majesty of a god.
Scarlett Johansson, famous for her role as Black Widow, is currently considered the 'representative of sexy voices', but Since childhood, her characteristic “husky low voice” has been a major disadvantage for her as an actress. This is the industry stereotype she changed.
- The barrier called ‘screen tone’ created by the recording environment
In the past, TVs didn't have the same great sound quality as they do today. Depending on the environment, there could be a lot of noise mixed in. CRT TVs and speakers of the past had weak mid- and low-range sound reproduction, and high-range sounds tended to be more audible to the ears .

If you imagine the distinctive sound of radio waves from the old days, you'll get the idea.

In order to “get it across” in such a poor recording environment, there was a preferred vocal tone .
Especially for voice actors, clear, mid-high tones are often given priority in casting criteria. The word that comes out here The word is “screen tone.” In Japan, it is also called ego-e (laughter tone) . It is a word derived from the characteristic of high-pitched laughter being able to spread far and accurately even in a noisy environment.
As for this ordinary voice and screen tone, it was expressed very well in the anime “Flowers Bloom. Like Asura”. If you're curious, check it out. It's an anime that you won't regret your time watching. (It's a shame though...)
In Japanese voice acting lectures or columns, the statement that "in the past, high notes were required to be heard well on television or radio" often appears.
Similar stories have appeared in NHK-related documentaries and interviews with voice acting schools. However, please note that this is only a 'common belief' and not a thesis-level verification!
Even in the days of radio dramas, high notes were preferred for similar reasons, and in the theater and broadcasting fields, the concept of a 'voice that is good for the microphone' is old. They even practice their vocalization separately for this purpose.
The stereotypes of “a recording environment that prefers high tones” and “a character pioneered by a good actor” mentioned above are combined. The early industry centered on high tones > fixed as a stereotype > created a trend that has led to high tones becoming mainstream even to this day .

So, even in the voice acting industry, you may feel that people with "low-mid voices" do not have particularly low-mid voices compared to the average person.

This tendency is particularly severe in classic anime. If you look at old anime, even the male characters have high-pitched, clear voices like cuckoos. (including female voice actors playing male characters) are common. This continues to this day.

Korea is a bit less so, but when you listen to the Japanese voices of the same main characters (One Piece, Naruto, Shin-chan, etc.),

You can hear them using high notes that are, to put it bluntly, almost frivolous.

According to voice actor Kang Su-jin, the Korean version of One Piece Luffy was played by a female voice actor (Mayumi Tanaka). There were times when the response was not good because of that discrepancy in tone…
- The stereotype that was created that way is “character = personality = cliché of voice tone”
Now, in the animation industry where fast development in one cour is preferred, this stereotype has started to be used in reverse . In order to quickly convey the character to the viewers , we used intuitive understanding through voice type rather than persuasion through acting .
So these days, at the animation planning stage, character settings → personality → voice image are already tied together and delivered.
At the scenario + voice actor audition stage, the direction of “this kind of voice” is set, and the tone is often the first thing to be judged.
Example: Active character → fast, high-pitched, bright voice. Cool character → low, slow voice, etc.
Because this formula has become established, there is a phenomenon in the Japanese voice acting industry where voice actors are trained to match their vocal range and style to the needs of the character. Voice actors tend to become fixated on characters that fit their voices , rather than the acting or challenge they want to do.
The current Japanese voice acting industry is one where people are chosen for characters based on their voice tone rather than their great acting.
- The voice of a fat person, a typical victim of stereotypes
Surprisingly, there is no “fat” in the vocal cord muscles.
The difference between acting as a “fat character” and the voice of an actual “fat person” is what exemplifies this classic stereotype. Voice actors who play fat characters often have dull, low-pitched, muffled voices.
However, there are no fat cells in the vocal cord muscles that produce sound. In other words, your basic voice will not change as you gain weight = as fat accumulates .
The voice follows the rules of wind instruments in the way it produces its sounds. Wind instruments with long tubes produce high notes (piccolo, etc.) and the longer and thicker the tubes, the lower the notes they produce.
Even on the same saxophone, the high notes (soprano) have a short, thin tube, while the low notes (baritone) have a long, thick tube.
So, rather , if the width of the vocal cords and larynx is narrowed (thin tube) by fat, the voice is more likely to move to the high range rather than the low range . As mentioned above, there are no fat cells in the vocal cords, so gaining a little weight will not change your voice. For your voice to change, you have to be extremely obese to the point where the surrounding fat cells narrow and affect your larynx.
“Big World” is a comedy show with a comical premise that big people have power.

If you close your eyes and listen carefully, you can clearly hear the rough 'high tone' .

However, in this case, it is more correct to view it as “voice for acting (communication) = screen tone” rather than a high tone due to obesity.
- Voice has nothing to do with personality. I've come back to this many times, but the bottom line is this: We often associate a character's personality with their voice, but Most of them are products of “stereotypes” created by prejudice . Prejudice created by actors who portray the character’s personality so well.
A person's personality, appearance, or voice have virtually no relevance. It's just a result of years of experience and learning from the media.
The current industry is using these stereotypes in reverse, Someday, if given the ability or opportunity, it would be great if someone appeared who could break these stereotypes.
+ Bonus: XX is a good language for expressing emotions .
People who watch a lot of Japanese anime and some others often say this. “Japanese is a language that consists solely of spoken pronunciation, making it ideal for expressing emotions.” (Actually, there are fewer spit-out pronunciations in Japanese - strong stop sounds (e.g. [p], [t], [k], etc.))
People who watch a lot of hand dramas often say this. “Thanks to its numerous vowels, Korean is excellent at expressing a wide range of emotions.”
People who prefer Chinese dramas often say this. “Thanks to the strong and clear pronunciation of Chinese, it is excellent at conveying emotions clearly.”
For those who like mid... “Because of its structural intuitiveness, English is a language that conveys emotions....”
This element is also a regular topic of nationalistic broadcasts in each country. Even some experts believe and quote this. But of course, this is a widely spread misconception .
This is simply because the actor (voice actor) who performed that scene did a really good job of transcending language . There is no such thing as a language specifically designed to express emotions.
For example, if we say a really touching line ourselves, we can immediately tell that the atmosphere is completely lost. If it were a language specialized in expressing emotions, then emotions should be conveyed no matter who speaks it, but that's not the case at all.
There was a time when I, too, watched anime as a child and believed what they said about Japanese being good for expressing emotions. It's just that the voice actor did a really good job...
It is a common topic among patriots, but from an objective point of view, it is a claim without any basis.
As we were talking about voice actors, I remembered something I had been writing for a while, so I quickly finished it and uploaded it. Because of this, please be generous and understand any parts that are a bit illogical or contain frequent typos ^^

Details