The Product Wasn't the Voice. It Was the Permission.

Ryan McDonald · · 7 min read · J-Pop Drop

Put this on while you read

The Product Wasn't the Voice. It Was the Permission.
Photo by MIKI Yoshihito, CC BY 2.0, via Wikimedia Commons

August 22, 2009. Animelo Summer Live at Saitama Super Arena. Thirty thousand square meters of venue, 25,000 people packed in, waiting for a two-song set from an artist who had never spoken a word in public, had no biography on record and was, technically, software.

The lights dropped. The crowd erupted before the first note.

“Miku Miku ni Shite Ageru” opened the set. Then “Black Rock Shooter.” The audience sang both songs from memory. They danced. They held glow sticks in the right colors. By every observable metric, this was a concert, with a performer and a crowd and a shared emotional experience. The performer was a projection on a screen. Nobody seemed to think that was strange.

It wasn’t strange. It was the logical outcome of a product decision Crypton Future Media had made two years earlier, one that had nothing to do with the voice engine.

The Brief That Built Miku

In 2007, Crypton hired an illustrator named KEI to design the character for their upcoming Vocaloid release. The brief was minimal: draw an android, use the Yamaha synthesizer color scheme, make it feel futuristic. That was it. No character history. No backstory to illustrate. Just an android in blue-green.

KEI added the pigtails himself. He liked how they filled out the negative space in the design. He referenced Yamaha’s DX-100 and DX-7 keyboard models in the skirt patterns. The floating ribbon restraints on the twin tails were his invention: a “special material” that held the look together. The design took the minimal brief and built something specific enough to be loved.

Why a character at all

Crypton’s competitors in the Vocaloid market, Zero-G (the company behind LEON, LOLA and MIRIAM), launched in 2004 with pure software tools. No character. No face. No name the user could attach to. Those products failed to find a mainstream audience. Crypton watched that and drew the obvious conclusion: voice synthesis technology alone wasn’t enough. You needed something people could care about before they opened the DAW.

But giving the voice a face was only half the decision. The other half is what actually matters.

Crypton gave Miku no backstory.

Not a thin backstory, not a vague paragraph in the manual. Nothing. The official documentation listed her height (158cm), weight (42kg), age (16) and favorite genre (denpa songs and techno pop). No parents, no history, no personality beyond what the design implied.

Hiroyuki Itoh, Crypton’s CEO, later described the logic: they wanted to “enable creators to tell their own tales without a predefined direction.” If Miku had a story, that story would constrain what creators could do with her. The absence of canon was the feature.

What the Blank Canvas Did

On December 3, 2007, Crypton launched Piapro.net alongside the Miku release. Piapro was a UGC platform built specifically for fan-created content: songs, illustrations, stories. The Piapro Character License gave anyone legal permission to create and distribute non-commercial works using Miku’s likeness. Not “we’ll probably tolerate this.” Actual, documented permission, written to be understandable by a teenager.

Nico Nico Dōga was already running. It became the distribution layer: producers uploaded Miku tracks, audiences responded, tracks hit 100,000 plays and earned Hall of Fame status. By July 2008, Crypton had sold 40,000 units of Hatsune Miku at an average of 300 units per week. A voice synthesizer.

The tracks spreading across Nico Nico weren’t just covers or demo songs. Producers were building entirely new aesthetics using Miku’s voice: hard electronic, cutesy pop, aggressive punk, ambient. Because she had no defined personality, she could hold all of it. Every producer’s Miku was a different Miku. The audience developed loyalty to specific producers, to specific interpretations, to the character-as-medium rather than the character-as-fixed-entity.

Backstage Pass ■ Access Granted

The Three Permission Layers

Legal permission. The Piapro Character License gave creators documented rights before they asked for them. Most companies in this position would have waited to see if fan creation happened and then decided whether to tolerate it. Crypton issued the license at launch. Preemptive permission changes what people make: when you know you’re allowed, you invest more.

Creative permission. No backstory meant no canon to violate. A character with a defined history creates anxiety in the creator: am I doing this right, does this fit, will the fandom come for me? Miku had no history to violate. Every interpretation was equally valid. That’s not emptiness. That’s the most generous thing a product team can do for the people who use their work.

Cultural permission. Nico Nico’s Hall of Fame threshold (100,000 plays) gave creators an honest external signal independent of label opinion or critical approval. Make something. Post it. Find out. The platform’s architecture meant that success was legible without gatekeepers deciding what success meant.

July 2, 2009

Six weeks before the Saitama concert, Sega shipped Hatsune Miku: Project DIVA for PSP. A rhythm game built around Miku, using songs the Nico Nico community had made. The game used songs from Nico Nico, ones producers had uploaded to the fan platform that Crypton’s own license had enabled.

The community created the content. Sega packaged it. Crypton collected on the character rights they’d built. By 2012, the franchise had sold 1 million copies. By 2014, 4.5 million. By 2018, 6 million.

The voice engine was never the business. The permission layer was.

August 22, 2009

So when 25,000 people sang along to a projection at Saitama, they weren’t suspending disbelief. They were doing what Crypton had built the whole system to let them do: taking ownership of something that was explicitly designed to be owned. They knew “Miku Miku ni Shite Ageru” because a producer on Nico Nico had made it and 100,000 people had played it before any label touched it. They knew “Black Rock Shooter” the same way.

Miku hadn’t toured. She hadn’t done press. There was no PR machine behind her. The audience was there because the permission layer had spent two years giving people reasons to care, one track at a time.

By 2011, Los Angeles. By the time she played Wembley Arena, the crowd was 8,000 people who had never needed a biography to decide they were fans.

LEON and LOLA

Right now, there are dozens of AI music tools in market. Most of them are extraordinary. The voice synthesis is better than anything Yamaha shipped in 2007. The interfaces are cleaner. The output is indistinguishable from human performance in some contexts.

Most of them will fail for the same reason LEON and LOLA failed: they shipped the voice engine.

What matters is the community that forms around the tool, the legal framework that lets people publish and profit from what they make, the deliberate absence of a defined use case so that users fill in the story themselves. Crypton didn’t build a fan base. They built the conditions for a fan base to build itself. Itoh’s description of Miku as “an instrument” was precise: you don’t need the manufacturer’s permission to write a song with a piano. That’s the whole point.

The AI music builders who figure this out won’t be the ones with the best synthesis. They’ll be the ones who understand that the output is just the beginning. The question isn’t what the tool can make. The question is what you’re willing to let people do with what the tool makes.

Miku turned 16 in 2007 and hasn’t aged since. The catalog keeps growing. The backstory she never had keeps getting written, one track at a time, by people who own it because Crypton handed them the title before they asked.

That’s a product decision.