Swery removes gen-AI voices from Hotel Barcelona, reworking Mixed after backlash

Hidetaka "Swery" Suehiro and White Owl have substantially reworked Hotel Barcelona, and the headline change is blunt: they removed gen-AI voices from the game called Mixed, after months of updates and a first release that failed to fully land.

Eurogamer frames the arc as a turnaround. When Hotel Barcelona launched last year, it did not impress the outlet. But as the months passed, and as updates piled up, Swery and White Owl appear to have won over more opinion. The gen-AI voice removal is the clearest “we heard you” signal in that shift, and it explains why this feels like more than just a patch note story. It is about whether the game’s ambition survives contact with the audience.

To understand why this matters to decision-makers, zoom out for a second. In games, the early reception is often destiny. Not because players are cruel, but because distribution algorithms, marketing budgets, and internal momentum tend to follow initial metrics like wishlists, refund intent, reviews, and word of mouth. A creator-led studio can recover, but the recovery has a cost: it must spend time and engineering capacity fixing the specific problems that created the initial skepticism.

That is why a project like Hotel Barcelona is so interesting from a board and leadership perspective. Swery is not a typical publisher-first brand. He is known for oddball, auteur energy, which can be a superpower for differentiation. It can also create risk when a new audience is not emotionally prepared for the experiment. The original reception, per Eurogamer, suggests the game did not immediately hit the right notes. “Charming and earnest” was part of the appeal, but that does not necessarily translate into broad excitement. If the game felt off in ways that players could point to, updates had to do more than smooth edges. They had to clarify intent.

Now, add gen-AI voices into the equation. Audio is not a cosmetic feature. It is the fastest path to immersion, and it is also where authenticity concerns tend to concentrate. Even when AI-assisted workflows are used within legal boundaries, players tend to interpret voice synthesis as a proxy for broader questions: Are real performers being replaced? Is a studio cutting corners? Is the creative team trusting the audience to accept something “generated” instead of performed?

Eurogamer’s update is therefore high stakes. Removing gen-AI voices is a change that is likely to satisfy multiple kinds of stakeholders at once. Players get a more human-perceived sound. Creators get a stronger alignment with craft and performance. And internally, the team reduces the risk of the conversation becoming dominated by ethics instead of gameplay. For executives, that matters because ethics-driven controversy can turn into a persistent drag on discovery, even after fixes.

There is also a broader regulatory and policy backdrop executives will recognize. Across tech and media, governments have been steadily tightening how AI content is disclosed, labeled, and monetized. Requirements can vary by jurisdiction, and enforcement timelines can be uneven, but the direction of travel is consistent: more scrutiny, more documentation, and more accountability. Even if a specific game change is driven by player feedback rather than a regulator, leadership teams have learned the hard way that customer sentiment and regulatory pressure can converge. When they do, the companies that already adjusted systems and workflows have a smoother landing.

So what exactly does “substantially reworking” Mixed imply? In practical terms, it signals that Swery and White Owl did not treat the issue as a minor tweak. They treated it as something that needed real structural change, and then kept iterating long enough for opinion to visibly shift. Eurogamer credits months of updates and characterizes the later period as a re-rating of the project. That is not just a creative win. It is an operational demonstration of persistence.

For other studios and investors, the second-order implication is straightforward but not easy: trust is a product feature. You can ship a distinctive game, and you can even ship imperfectly, but the recovery plan determines whether the audience eventually feels respected. A fast pivot on something as sensitive as voice can be a signal to both players and partners that the team will absorb feedback rather than defend missteps.

In other words, Hotel Barcelona is less a singular story about one roguelike and more a case study in what it takes to reverse early skepticism. Swery removing gen-AI voices from Mixed, as part of a broader rework, gives executives a template for how creator-led teams can course-correct without losing their identity. It also raises the bar for peers: if you want to innovate loudly, you need to be able to iterate loudly too, especially when player trust is on the line.