2月にクリプトン在社時代の仕事にスポットライトが当たったことを記していましたが。
以前やり取りをしていたからメッセージが届いて、少し長文の返信をしたのでした。
あらかじめ書いておくと、今どきの𝕏ユーザーがよだれを垂らしそうな対立の話題では全くなく、先方の興味に応えただけの内容です。
で、伏せるような内容でもないので、ここに転載しておこうと。ただ念のため先方のお名前は伏せておきます。
Hello Makou,
Long time no speak! I wanted to follow up with you after a long time to ask you about your ringtone work with Crypton Future Media and beyond.
Allegedly, you created hundreds of ringtones!! What was your process working on ringtones and what are some of your favorite ringtone compositions you’ve made over the years?
Thank you,
この返信として、その私視点での前夜後夜の話を次のように記しました。
Hi,
To avoid any misunderstanding, I’ll explain the background in a bit of detail.
I started working on ringtones around the year 2000. At that time, Crypton Future Media’s main business was importing overseas sampling CDs and sound effects libraries. One of the staff members proposed an idea to promote those products by converting the sounds into a format that could be played on mobile phones, which many people owned.
Back then, communication bandwidth was extremely limited, and in the beginning we had to compress everything into very tiny files—sometimes as small as 5KB. Because of this limitation, the focus shifted to how we could make sample materials sound interesting within that strict 5KB constraint.
At the time, it was common for ringtone services to sell arrangements of famous songs. However, because of the approach above, our content became quite different, and it started attracting attention from enthusiasts.
The sound chip used for ringtones at that time was made by YAMAHA. It could handle both FM synthesis and ADPCM-compressed audio at the same time, while the MIDI data itself had to remain extremely compact. By making full use of these features—and sometimes even exploiting bugs—we tried to create content that would surprise or amuse the users.
When Crypton Future Media launched a mobile website called NorthSound for J-Phone (later Vodafone and SoftBank), we initially needed a stock of about 100 tracks, and we had to add two or three new tracks every week. Eventually the project began attracting attention from both J-Phone and YAMAHA. Later, when a new site called Melodiclover was launched jointly with YAMAHA, we needed about 200 tracks in stock. There were four full-time staff members working on it, and we handled the workload at full speed.
So the main reason I created hundreds of ringtones is simply because that many were required. Another factor was the technical limitations. With the 5KB size restriction and the lightweight sound chips, each piece could be very short and didn’t need to reach the level of completeness expected from a full track. The trade-off, however, was that I had to keep generating new ideas constantly, almost without pause.
Both NorthSound and Melodiclover centered around “club music.” However, when I first started working on ringtones, I actually knew almost nothing about techno or house music. I had to study them intensively. At that time the internet didn’t have much useful information yet, so the only reliable sources were CDs and vinyl records. Most of my salary went into buying them.
Since the top priority was simply to produce enough tracks, I turned every idea that came to mind into a finished piece. In reality, aside from sleeping, I was almost constantly thinking about the next track to create—searching for ideas and preparing the next piece.
One of my personal favorites is a later piece called Life Style #591. I would say it became the starting point for the technique I later used when providing music for DJMax and other rhythm games—building tracks by incorporating various vocal sampling materials.
I may have mentioned this in a previous interview, but in a rural city like Sapporo it’s actually very difficult to find excellent singers, instrumentalists, or people who can write English lyrics. Because of that, when I’m asked to provide vocal tracks for games like DJMax or other rhythm games, it’s honestly quite challenging.
In recent years, the spread of AI has begun to reduce that disadvantage somewhat. However, there are still many people who criticize you the moment they hear the word “AI,” so I sometimes wonder when we will reach an era where we can simply enjoy making music freely again.
Best regards,
Makou
短期間でそれなりの曲数を揃えるのが大変だったという話、クラブミュージックのカテゴリーを知らないまま取り掛かったためYouTubeもまだない時代に資料収集にすごく費用がかかった話、あとは近年のAI狂想曲な状況に少しだけ触れた感じ。
当時と取り巻く状況が著しく違うので、上記エピソードがいま参考になるとは思わないのですが、それは私から見ての話でしかないので、もしかしたら参考になる方がおられるのかもしれない、ならばどうぞということで引用しました。
今どきのモノ作りにおいては、SONICWIREのブログの最新インタビュー記事が非常に参考になるので、その道を志す人は目を通すべき。
タイトルからするとAIに絞った話に見えますが、そうではなく、音楽で食ってくモデルケースとして踏まえておくべきことが綴られています。
ちなみに最近音楽生成AIに関する一部話題が𝕏（で私にとって最適化されたTL）を賑わせていましたが、私自身、楽曲制作の中で音楽生成AIを試したり（この間書いたようにAIの限界を感じたのでもう使ってない）、支援AIを取り扱う業務に携わったりしている身からすると、既に記したとおり、警戒、拒絶されてる一部の方が思うほど音楽生成AIは優れてません。自身の音楽スタイルを大事にしていればしているほど、生成AIは逆に効率が悪いですからね。
昔ながらの二項対立で言うならば、AIに使われるか、AIを使うかの後者にならない限り、使う意味は薄いよなというのが現在の自分の思うところ。