Reimagining DiDa —

An AI Music Experiment

The original Apple Watch DiDa campaign used a soft R&B track. I always felt the tick tock concept could be pushed further. Here's just trying what was in my head, by using Claude as the guide and Suno for music creation.

Original Apple Watch China campaign — DiDa

Claude: What do you feel is missing from the original DiDa track?

Me: Currently song feels too soft. The tick tock idea doesn't come through strongly enough. I want something with more tension but it still has to work with beautiful choreography.

Claude: What specific references come to mind?

Me: The intro of G-Dragon's Power — distinctive and playful. And the FKA Twigs track from the Apple HomePod commercial — that descending melody feels inevitable. The original is missing a memorable hook.

Claude: Try this prompt:

Electronic track. Mechanical tick tock pulse as the core rhythm — precise, relentless, slightly hypnotic. Layered arrangement, not thin. Playful but restless, like a clock that never stops. Modern production — somewhere between industrial electronic and contemporary K-pop. No obvious melody yet, just the rhythm building. Urban, driven, slightly suffocating beneath the surface. 45 seconds.

Demo01/K-pop created by Suno

Me: What if we try acapella? Human voices as the instrument.

Claude: Strong direction — acapella forces the DiDa to be the only thing. Nothing to hide behind. Try this prompt:

Acapella track. Human voices creating a clock rhythm — mouths, tongue clicks, breath — spelling out Di Da Di Da clearly and repeatedly throughout. Like a human clock. Builds from a single voice to a group of voices layering on top of each other. No instruments, no electronic production. Just voices as the instrument. Playful, hypnotic, slightly obsessive. The Di Da never stops — it's the foundation everything else is built on. BPM 104. 60 seconds.

Demo02/Acapella created by Suno

Me: What about a marching direction? Collective, intense, relentless?

Claude: A marching song with DiDa as the pulse feels completely different from what exists. It captures the 996 tension in a way the original soft R&B track never could. Competitive, relentless, slightly militaristic — which is actually how that work culture feels from the inside.

Acapella marching chant. Human voices chanting Di Da Di Da as a driving, relentless pulse — like a military march or competitive sports chant. Intense, disciplined, slightly overwhelming. Multiple voices layering — some chanting the rhythm, some building on top. No instruments. The feeling of a crowd moving in lockstep, no room to stop or slow down. Tension building throughout with no release. Like workers marching through their day. BPM 108. 60 seconds. The Di Da must be loud, clear and unmistakable throughout.

Demo03/Marching created by Suno

Me: Too grand, too military. Cannot fit the lifestyle and the soft choreography of the woman lead. Feels heavy. What about strings then? Could that create urgency without aggression?

Claude: Philip Glass influenced — urgency through relentlessness rather than volume.

Me: Ref feels dated for young adults. Loses the beat. Choreography has nothing to lock onto.

Claude: Focus on acapella — it's been the strongest direction so far.

Learnings:

Music is so technical and hard to describe. Claude helped translate feelings into terms that Suno could actually understand. That translation layer is where AI genuinely added value — turning a vague creative instinct into something specific enough to generate. But I also learned the limits. This works best when you already have a clear vision and a thorough understanding of what you want. Right now AI music is more about visualising a feeling — creating a demo for presentation, or for projects that don't require a highly polished final track. For anything requiring genuine craft and precision, you still need a real composer.

Knowing that is useful too.