ALEXVNDER BENDS THE FUTURE OF AI MUSIC WITH ‘WE ARE EACH OTHERS’ & HONEST BREAKDOWN OF THE NEW MUSIC MODEL

Here is a deep‑dive feature/opinion essay on the Suno V5 drop, what it might mean for artists, and how I used it under my producer alias ALEXVNDER (Dustin Hollywood) on WE ARE EACH OTHERS.

---

SONOCRACY RISING: WHAT SUNO V5 MEANS FOR ARTISTS, FOR MUSIC, FOR ARTIST LIKE ME..

When I first got the beta access to Suno v5 (rolled out September 23, 2025) I felt a hard crack in the facade of what “AI music” has always promised, and often failed to deliver. This update is being billed by Suno as “more immersive audio, authentic vocals, and unprecedented creative control.” (ForkLog) It’s a bold claim, but bold claims are what shake up paradigms.

Instead of writing a dry spec rundown, I want to talk about what this means in real terms, what artists can do with it now (and soon), and how I used it to make WE ARE EACH OTHERS. Because that’s where theory meets flesh—and where this tool becomes something more than “just tech.”

---

THE LEAP V5 CLAIMS

Before I dissect its possibilities, here’s what Suno and observers are saying V5 brings to the table:

• Realistic, expressive vocals. The era of robotic singing with flat emotion seems to be receding. The new model claims to close that gap.
• Better audio quality and spatial depth. The mix, stereo field, sense of “space” is reported to be richer.
• Finer control over structure, genre blending, and prompt compliance. The model is supposed to obey more precisely — e.g. length, transitions, instrumentation, layering harmonies, etc.
• Genre fusion and recognition. Where earlier versions might struggle to merge styles, v5 reportedly handles blending better.
• Rollout to Pro / Premier users. It’s not universally open yet; it’s in beta for paid tiers.

Suno’s own benchmark (ELO-based) puts v5 above v4.5 and earlier versions in their internal ranking.

Critics and users, however, already have feedback: some say v5 is an improvement especially in melody and emotion, others argue for mixed results and leftover artifacts in instrumentation or genre extremes. On forums, I saw users claim:

“Better mixed? Hell no.”
“Works better in some genres — but harsh vocals or metal still sound odd.” (Reddit)

So it’s not perfect. But it’s a leap.

---

WHAT ARTISTS CAN DO WITH V5 (NOW AND BEYOND)

I’ll break this into three dimensions: composition, performance, and collaboration.

Composition: idea-to-sketch and sonic prototyping

For me, this is the killer app. Before, generating a rough song sketch with voice, chords, instrumentation, something you can immediately audition and iterate, has been laborious. V5 lets you lean harder into what I call “creative friction removal.” You toss in a prompt: genre cues, mood adjectives, instrumentation, vocal style, section lengths, transitions, and you get back a version you can re-prompt or morph. With the finer control, you can say: “I want 8 bars of ambient synth, then 4 bars break with a vocal solo, then a drop, then a harmony chorus.” The model is said to obey that with higher fidelity. (Aibase)

For producers, that means the distance from concept to demo collapses. You’re not always building from zero. You can “jam” with the model, fight with it, sculpt with it, then bring it into a DAW and treat it as source material.

Performance: synthetic vocals, harmonies, textures

The push to more authentic, expressive vocals is a frontier I’ve been watching closely. If v5 can generate vocals that carry emotion, nuance, performance inflections, then artists can now co-write with the AI voice. You might treat it as an extra singer, a co‑vocalist with its own tone and timbre. You can layer harmonies, counterpoints, ghost vocals, doubling textures, all from prompts or controlled variation.

That opens doors: singing in styles outside your vocal range, sketching out backing vocals without needing extra singers, sonic experimentation in micro-genres. For artists who produce and vocal, this is like having an ever-present “vocal partner” in your pocket.

Collaboration, rework & remix modes

One thing I see v5 enabling is a new type of hybrid collaboration: human + AI + human remix. Let me illustrate:

• You feed in your own vocals or stems and ask v5 to generate complementary parts (bass, synth, percussion) around your voice as if it were part of the model’s prompt.
• A co-writer or collaborator can take your v5‑generated version and remix, re-prompt, re-structure. Because v5 gives more control, the remix forks will be less chaotic and more musical.
• For licensing / scoring, you can generate alternate versions (instrumental, vocal stems, ambient versions) almost instantly, giving yourself more options to place songs in film, games, ads.

Looking ahead, the control layer might evolve into a lightweight “Suno Studio” interface (rumored to arrive soon) that lets you see stems, adjust balances, morph segments, pivot paths.

Pitfalls, ethics, and legal pressure

I have to speak honesty: this new power comes with acceleration of tensions. Suno itself is under legal scrutiny. The RIAA and major labels have sued Suno and Udio over alleged copyright infringement, claiming their models used copyrighted recordings at scale without license. (Reuters)

While models like Suno argue their use is transformative, critics say “training on copyrighted work to then compete with it” contradicts fair use boundaries. (The Verge)

On the technical side, the SONICS research paper shows that synthetic songs (end‑to‑end generated) can be distinguished from human-made ones by models trained on temporal and structural cues. That suggests generative systems must grow more sophisticated, or risk being fingerprinted, flagged, or regulated.

Artists will have to choose their relationship with AI carefully: as a tool, not a replacement; with full transparency and perhaps negotiation around attribution, licensing, revenue splits. The line between “model as collaborator” and “model as ghostwriter” will blur, but we have to guide it.

---

WE ARE EACH OTHERS: MY CREATIVE JOURNEY WITH V5

Now let me talk about what I actually did with it, how I used it, where it succeeded, where I had to step in and correct or rework, and what the experience felt like.

I’m ALEXVNDER in the music producer world, Dustin Hollywood in my world. This track WE ARE EACH OTHERS is my first full song made with V5 as a creative co‑pilot.

Prompting & iterations

I began with a high-level prompt: “Melancholic mid-tempo electronic pop, moody synth pads, warm analog bass, female lead voice with attached echo and harmony, structure: intro – verse – pre‑chorus – chorus – bridge – final chorus.”

The first draft the model gave me had a good skeleton but needed refinement: vocal enunciation was sometimes mushy, transitions slightly abrupt, the drums generic. I fed back adjustments: “clean up vowel articulation, soften snare resonance, insert a break before final chorus, add ghost harmony under chorus in minor third.” Over 4–5 cycles I got a version I felt had spirit and melodic contour I wanted.

Vocal work

The lead vocal in the model output felt expressive. It wasn’t perfect: sometimes sibilance or vowel consistency would slip. I exported the generated vocal as a stem, ran it through EQ, multiband compression, de‑esser, and subtle saturation. Then I layered my own human vocal in a few lines to bring “human weight,” blending AI + me. The harmony parts (generated) I kept mostly intact, only pruning occasional odd intervals or resonances.

Arrangement & transition shaping

In the AI version, the drop to bridge was sudden; I smoothed it by injecting an ambient pad fill, morphing the prompt, then stitching with crossfade and automation. For the ending, I faded into ambient reverb tails, letting layers dissolve. Because v5 gave good control over stems or sections (relatively speaking), I could cut, duplicate, re-prompt small segments, reinsert them in context.

Final mixing & mastering

I treated the AI output as raw material. The mix chain: bussing drums, mids, mids-high, vocal group, then stereo bus. I used dynamic EQ, bus compression, harmonic exciters (sparingly), and subtle stereo widening. The mastering pass was gentle, preserving dynamics. The goal was: don’t force the AI result to be something it’s not—enhance, correct, glue.

When I listened back, WE ARE EACH OTHERS felt like something born of a collaboration. You hear the temperature of human choice, the grip of human correction, and in parts, the novelty of the AI voice humming in spaces I might never have tried alone.

From the audience reaction so far (on my X post) there’s curiosity: “Is this human? Is this AI?” That ambiguity is intentional. The message of the song—interdependence, overlapping selves, mirrors its creation method.

---

WHERE THIS COULD GO: FUTURE SCENARIOS

These next moves are guesses, but I feel many are close.

1. Interactive, real-time co‑creation tools
   In a few versions, I expect Suno (or clones) to support live jam mode: you sing or hum, it augments in real time; you play chords, it layers. You’re in a sliding scale between “you play everything” and “the AI plays everything.”
2. Stem-level editing interface / modular control
   Rather than treat output as a monolith, future versions will give you editable stems, you’ll tweak levels, mute instruments, morph a bar or two, discard half the output. This “Suno Studio” vision is likely coming.
3. Conditional fine-tuning by artist
   Artists might be able to fine-tune a version for their own voice signature, timbre, style, personalizing the model to their sound across songs. Because v5 is already closer emotionally, this fine-tuning would push it further.
4. Integration into DAWs, plugin embedding
   A VST / plugin version of Suno v5 (or v6) will bring the model directly into Ableton, FL, Logic. You’ll prompt inside your session, generate stems in place, automate model changes.
5. Licensing economy, co‑credit, revenue sharing, regulation
   As the power grows, the legal terrain will demand evolving frameworks: model credits, shared royalties, AI “featuring” credits, regulators ensuring artists’ rights. Some artists may refuse to use AI models that don’t respect attribution.
6. Cross‑modal creativity expansion
   Since musical models overlap with text, voice, image, I see multi‑modal creators: generate the sonic bed, then feed the mood into a visual AI to make artwork or motion. The idea: a unified creative session: text → image → music → video, all emergent and linked.
7. Authenticity detection arms race
   As we saw in the SONICS paper, synthetic detection models (is this generated or human) will grow more refined. The creators of generative models will have to evade detection or work with regulators.

---

WHY THIS MATTERS, FOR ARTISTS AND FOR ME

For artists, the arrival of a more powerful model is not a threat (if you choose properly); it is a hammer we can pick up or reject. It democratizes access to sophisticated musical prototyping. It demands we redefine boundaries: what is composition, what is performance, what is collaboration.

For me, using v5 on WE ARE EACH OTHERS helped me traverse a creative territory I might never have spontaneously discovered, textures and voice ideas that felt partially alien, partially human. The synergy between me and the AI forced me to remain alert, reactive, critical, playful.

If v5 is a crack in the barrier between what’s possible and what’s typical, then I’m already standing halfway through it, turning back to glimpse what’s behind. To the artists reading this: try it now, fight with it, reflect on what choices you insist on keeping, and decide whether the model is your instrument—or your co‑author.

www.suno.com