
PC memory prices are soaring due to AI companies buying up memory.
Eventually, the AI bubble will burst and prices will fall, but no one knows when that will happen.
Various theories are floating around, but the general consensus seems to be that PC (and smartphone) prices will remain high at least until the end of next year.
PCs are essential tools for my creative work, so I won't hesitate to buy one if I need one, even if the price goes up a few percent. Honestly, it's troubling when they're in short supply and difficult to obtain.
I can't stand a life without creative work for an extended period of time.
The computer I currently use primarily for creative work is a custom-built PC I built about five years ago, and it's been working flawlessly so far, but I leave it on all the time, so some of the components are about to reach the end of their lifespan.
I do have a laptop, but it's difficult to create solely on a laptop for long periods of time.
So I bought the latest M4 Mac Mini.
Well, my main PC has been freezing frequently when using heavy audio files lately, and Windows 11 updates have been plaguing me, so I've been missing my Mac, and I finally decided to go for it.
The price was a shocking 90,000 yen.
I didn't take out a loan.
As expected of Apple, they had ample inventory, so I was able to purchase it at the regular price without any memory issues.
Compared to my previous Mac Mini (Late 2012), it's significantly narrower in width and depth, and slightly thicker.
The minimum configuration is a 256GB SSD and 16GB of memory.
Mac pricing is a bit unusual, with prices rising exponentially as you add more storage and memory.
The cheapest Mac Mini is like economy class on an airplane.
I feel like I'm being spoiled by a wealthy person who can afford a high-spec Mac.
I would ideally like to have a 2TB SSD and 32GB of memory, but this configuration cost a whopping 270,000 yen lol.
That's definitely way over my budget.
I've been making do with 16GB of RAM until now, and I can easily add an external SSD, so I think I'll be fine.
My desk will be cluttered with cables, but that's irrelevant; cost-effectiveness is everything.
Thanks to making the purchase early, I was able to get both the external SSD and HDD for a much cheaper price than I am now.
I haven't had much time to test the performance with DTM yet, but even when I loaded all the instruments from BBC Symphony Orchestra Core and played them, there was no skipping at all.
On my previous PC, I would sometimes experience skipping even when the buffer size was set to maximum, so this is a much smoother experience.
I think I'll be able to do DTM comfortably over the New Year holidays.
And that's it!
The human ear is quite imperfect, and it seems we have a tendency to mistake louder sounds for better sounds.
That's why the majority of music out there isn't the exact sound as it was recorded. Rather, it's compressed using compressors and limiters, and processed to sound as loud as possible within the volume limits of the media.
...but that was back in the days of taiko, when CDs were still the norm.
With the rise of streaming services that came standard with an automatic volume adjustment feature called loudness normalization, the loudness wars came to an end as the carefully layered compressors and limiters were applied to avoid distorting the sound, nullifying the sonic pressure that had been so well-earned.
Decibels (dB) are the unit of measurement for the actual loudness of sound, but there is also a separate value called LUFS, which is based on human hearing.
This is what we call sound pressure or loudness.
Even if the dB value is low, the higher the LUFS value, the higher the sound pressure and the louder it will actually sound.
LUFS also comes in three types, determined by the length of measurement time: Momentary, Short, and Integrated. The specific measurement times are Momentary 0.4 seconds, Short 3 seconds, and Integrated for the entire song.
I looked into which LUFS value each streaming service uses as the basis for loudness normalization, and found that the official Niconico page listed "Integrated."
YouTube's official recommended LUFS is apparently -14, but unlike Niconico, there is no specific explanation page to be found, so the details are unknown.
Subscription music streaming services have not yet released songs, so they were outside the scope of this verification, but I think I saw -14 LUFS (integrated) listed on the official Spotify page.
If we were to take the official information at face value, -14 LUFS would be the optimal solution, but a while back, I conducted a survey on X asking my DTMer followers what LUFS value they actually upload their songs at, and quite a few people said it was around -8 to 10 LUFS, so I decided to put it to the test.
First, I prepared sample data of the same song with LUFS values (Integrated) set to -7, -10, and -14.
I uploaded these to Nico Nico Douga and YouTube, played and recorded them on my PC, and then measured the LUFS value again.
The sample was an original song I just finished.
It's for Yuudachi-P's end-of-the-world compilation, so it's scheduled to be released at the end of the year.
-7 LUFS test data waveform (screenshot of WAV export using Reaper).
-10 LUFS test data waveform.
-14 LUFS test data waveform.
First, YouTube.
Original data: -7 LUFS
The remeasured value was intentionally set low to avoid being cut by the limiter.
Original data: -10 LUFS
Original data: -14 LUFS
Test results: YouTube is barely affected by loudness normalization.
However, when I checked YouTube's settings, the automatic volume setting was turned off and grayed out.
Is this a feature exclusive to YouTube Premium?
Next, Nico Nico.
Here, the automatic volume adjustment was switchable (perhaps because I'm a Premium member?).
First, I tested it with automatic volume adjustment turned on.
Original data -7 LUFS
Original data -10 LUFS
Original data -14 LUFS
Test result: The LUFS value is adjusted to the same value regardless of the original data's loudness.
Next, try turning off automatic volume control.
Original data -7 LUFS
Original data -10 LUFS
Original data -14 LUFS
Test result: No effect of loudness normalization (obviously).
Conclusion.
The optimal loudness value varies depending on whether you target users who turn automatic volume control on or off.
I wonder how many people actually listen with automatic volume control turned off.
I don't listen to music at high volumes, so I usually keep automatic volume control on.
Well, I thought it would be best to strike a balance and post at around -10 LUFS.
In other words, it's pretty much the same as before.
And that's it!
Hello! This is OkonomiP, also known as ENIN.
Since Makino Sumire's birthday celebration was on September 15th, I posted my work for the occasion.
This time, I tried my hand at 2D animation for the first time.
My previous 2D music videos were made by somehow animating a single image, but since the image itself wasn't moving, I had to forcibly create movement using text animation, effects, and free video clips, so I really wanted to try animating the characters themselves.
I've been using Clip Studio Paint (formerly known as Manga Studio) for drawing for a long time, and it also has a function for creating animations, so I used that this time.
Being able to see the previous and next frames while drawing, and being able to test the animation immediately, makes it much more efficient than simply preparing many still images and connecting them with video editing software.
I intended to post it with plenty of time to spare, but ended up posting it at 11:59 PM on Makino Sumire's birthday—a ridiculously last-minute post! I really need to improve my scheduling skills.
Regarding the song, the inspiration came from my real-life experience when I was traveling by bicycle across America. I was approached by a woman living in a small village in Kansas, who offered me a place to stay for the night.
She had what looked like a hunting rifle displayed in her house, and although I didn't actually see her use it, I imagined her riding a horse across the prairie while holding a rifle.
The wind-like sound effect at the beginning is from a video I recorded in the prairies of Kansas. It was tough pedaling against the wind across the endless prairie (lol).
For this project, I didn't have much time, so I used a template from my previous project, Maki-chan's Okonomiyaki Dance, and repurposed it.
SERUM is quite old though...
Maki-chan has gotten a lot more powerful after updating to SV2.
New vocal styles like Whisper, Cute, and Grunge were added, allowing her to do everything from an adult-sounding voice to a loli voice, and even a neutral, cool-sounding voice.
I spent way more time on the music video than I expected, so I had to greatly simplify the mixing and mastering. Now that I listen to it, the sound feels a bit rough, which is a shame.
But I think being able to finish things by the deadline is also a skill, so I won't re-post it; I'll just learn from this for next time.
On the other hand, if I had spent more time on mixing and mastering to perfectly finish the song, the music video probably wouldn't have been finished. So, working on both the song and video simultaneously and balancing them was probably the right approach.
If I had more time to work on it, I wouldn't have needed to do that, but if I have too much time, I end up working on other songs and still finish it at the last minute, so there's no perfect solution.
I've been creating things for decades with a fatal flaw in my ability to plan and execute, and I probably will continue like this until I die, but I'm going to keep pushing forward regardless.
I want to keep moving forward, even if I'm lost.
And that's it!
I write about this frequently on X and Bluesky, but I think it would be tiring to just spout my thoughts on social media, so I'll just scribble them down here, which is my diary.
First of all, I'm not what you would call anti-AI.
I write "No AI learning" on this website and in my social media profiles, but I mean please don't use my work for any purpose other than viewing it without my knowledge, which is the same as no unauthorized use.
I would like AI itself to develop as a tool to make life and creative activities more convenient, and if it can streamline the creative process and give form to ideas more quickly, I would use it too.
I have tried image generation AI like Mid Journey, music generation AI like SUNO AI, and ChatGPT, and thought they could be used to build the foundation of my work and generate ideas. However, I felt that the source of their learning is not clear, and that they are created through unauthorized learning, and that they can reproduce specific styles like Ghibli art, so I don't use them now.
It's not that AI itself is bad, but rather that using copyrighted material without permission, whether it's images, music, or text, is wrong.
I'd rather use clean AI like Synthesizer V or Voisona, which have been properly put into practical use after paying and obtaining permission from the voice actors they trained on.
I don't understand Vocaloid tuning either.
This is purely my personal opinion and preference, so I have no intention of forcing others not to use generative AI.
Recently on X, an official city VTuber caused controversy by saying something along the lines of, "I'll keep my distance from businesses that use image-generating AI," but it's not okay to arbitrarily choose which businesses residents should support based on their personal opinions or likes and dislikes.
I wouldn't like it if the poster character of my town said, "I'll keep my distance from businesses that don't serve vegan or halal food," either, and personal opinions and the actions that accompany them should only be discussed after giving up the city's poster boy.
Furthermore, people who threaten arson, like in the case of Kurumazaki Shrine, are simply criminals.
But that's a digression.
I think there are all kinds of different types of generative AI users.
Some people simply enjoy playing with it as a toy, while others, like some Vocaloid producers, use it in music videos and other creative endeavors.
In particular, when it comes to illustration-generating AI, perhaps because I've been drawing since I was a child, I can somehow tell when someone is using AI, even if they don't publicly admit it.
It just doesn't feel like it was drawn by a human.
Sometimes the images are unnaturally neat, and sometimes they make mistakes that humans wouldn't.
Everyone has different thoughts and circumstances, and it's not prohibited by current law, so no one has the right to stop the use of generative AI itself.
Making and selling counterfeit products of famous creators, or distributing AI created by incorporating a voice actor's voice without permission, or misusing it for spreading rumors or scams, are a whole other legal issue.
It's just a matter of how you use it.
There may be some people who like the pictures created by AI, and there have been stories of winning entries in art contests being made by AI, so if you keep spinning the gacha machine and coming up with creative prompts, you might just create a masterpiece.
But...
As someone who creates both art and music myself, I want to say loud and clear above all else:
Creating things yourself is more fun!!!
That's it.
Drawing pictures is fun, and making music is fun.
Playing with generative AI might be fun, but I want you to experience the joy and happiness of creating something yourself, and the sense of accomplishment and fulfillment you feel when you complete it.
My life is supported by creativity.
I can't help but enjoy creating, and knowing that there are people who are pleased with my work makes me work hard even at laborious times.
It gives me the courage to survive in a world full of unpleasant things.
Even if I were reborn as an AI that could create even better works than I do now, I would continue to create my own works.
I would create characters that reflect my fetishes, draw pictures of my favorite idols that reflect my fetishes in every detail, and compose songs using my favorite rhythms, chords, tones, and even my favorite idol's voice.
It's just the best.
I wrote this article while listening to a new song I'm working on over and over again.
My songs are amazing, so please give them a listen on the music page of this site.
If you like them, please give them a thumbs up and subscribe to my channel. I'll cry tears of joy.
And that's it!
I recently learned that a system called the "Unmanaged Works Adjudication System" will be implemented next year.
For details, please see the announcement from the Agency for Cultural Affairs.
Unmanaged Works Adjudication System
Copyright Act Amendments at the 2023 Ordinary Diet Session
What is the Unmanaged Works Adjudication System?
To briefly explain what this system is, it allows you to freely use copyrighted works that are not managed by a copyright management organization like JASRAC and do not have explicit terms of use if you cannot contact the creator.
I was shocked to see a statement like this at the beginning of "4. Q&A on the Amended Law."
I mean, if you can't get permission, don't use it.
I've also seen talk of paying a 13,800 yen fee to the government when the ruling is made, but this isn't mentioned on the page above, so the details are unknown.
The Agency for Cultural Affairs says that as a benefit for creators, even if their work is used without their permission, they can receive compensation if they come forward.
The aim is to encourage the use of copyrighted works that copyright holders themselves were unaware of.
Honestly, this is going too far.
If creators want others to use their work, they should set up a contact point or publish guidelines, and users should also seek them out.
I hate the idea of people using their work for a fee arbitrarily decided by the government.
It seems that if the author's work is used in an undesirable way, the author can request a cancellation later, but even if it is possible to prohibit the use itself, it is nearly impossible to erase a work that has already been released to the public.
This can happen simply because someone hasn't replied to an email for 14 days.
In these days of rampant scams, it's only natural to be wary of emails from people you don't know, and it's possible that the email will end up in your spam folder and be forgotten.
The process of canceling a cancellation is sure to be a pain.
How can an unknown creator prove they are the author?
That said, this system does not apply to all creative works.
If the author simply makes the terms of use of their work clear on their website or social media profiles, it will no longer be covered by the system.
That's why I prohibit the unauthorized reproduction, use, and AI learning of all my works.
I have written this throughout this website and clearly stated it in my social media profiles.
I don't want to waste my remaining time and effort on unnecessary arguments; I would rather create if I had the time.
Please do not do so without permission, but if there is anyone who would like to use my work in something, please contact me via the contact form on this site or on social media and I will be happy to discuss it with you.
However, if I am unable to reply for some reason, please understand that this means no.
And that's it!