- Industry: Hospitality
Property: A 180-room independent hotel in Lisbon, Portugal, drawing roughly 60% of guests from outside the country
Challenge: Deliver consistent multilingual guest communication without expanding staff
Solution: PolyTalk, a self-hosted speech translation platform for live guest-staff conversations
Deployment: Self-hosted, to keep guest data on the hotel's own infrastructure
Departments involved: Front desk, concierge, housekeeping, dining, guest services
Outcome: Check-in conversations that used to take four minutes now run under ninety seconds for non-English-speaking guests, and housekeeping requests that once took two or three calls to resolve now typically close on the first one
At a glance:
- One workflow, six guest-facing departments, no separate tools for different situations
- No new hardware or IT overhaul, runs on infrastructure the hotel already controls
- Guest audio and conversation data never leave the property's own servers
- Full rollout from the front desk to five departments took three months
The Problem: Language Barriers Were Slowing Guest Service
A guest at the front desk switches to Spanish mid-question, asking about a late checkout. The associate helping her speaks only English and Portuguese. A beat later, she hears the question in English. She answers normally. The guest hears the reply in Spanish before she's finished her next sentence.
That moment happens dozens of times a week at hotels serving international guests, and for a long time, it was one of the most disruptive parts of the job.
The hotel sits a few minutes from Lisbon's waterfront and draws guests from across Europe, the Middle East, North America, and Asia. Most staff spoke English well. Not every guest did.
That gap showed up everywhere:
Check-in slowed whenever staff had to explain payment details or room policies through a translation app instead of speaking naturally.
Concierge requests paused while someone searched for a colleague who spoke the guest's language.
Housekeeping and maintenance requests sometimes took multiple conversations before the issue was understood.
In urgent situations, such as a medical concern or a lost passport, there was little time for typing messages into a phone.
None of this was a service failure. As the front desk manager put it during a review of the front desk workflow:
"Our people know how to take care of guests. They just couldn't always understand them."
That distinction mattered. The hotel didn't need more multilingual staff. It needed the staff it already had to communicate clearly, in real time, regardless of which language a guest spoke.
Why Translation Apps and Guest Messaging Weren't Enough
The hotel already used two tools: a guest messaging platform for confirmations and reminders, and phone-based translation apps for live conversations.
The messaging platform worked fine for what it was built for, sending information, not having a conversation. The translation apps were the real bottleneck. A guest would ask a question. A staff member would open an app, speak into it, wait, then read the translation aloud. A simple exchange that should take ten seconds took closer to a minute, and the back-and-forth made conversations feel transactional instead of personal.
The core issue wasn't a missing feature. It was a missing category of tool, something built for live, spoken conversation, not written messages or app-mediated exchanges.
How Real-Time Speech Translation Fits into Everyday Hotel Operations
The hotel first deployed PolyTalk at the front desk before expanding it across guest-facing departments. The workflow behind that late-checkout moment is straightforward:
Guest speaks → speech is translated in real time → staff hears the reply in their own language → staff responds → guest hears the reply in their own language.
No phones. No awkward pauses. No searching for a bilingual colleague. For both the guest and the employee, the interaction feels like a normal conversation rather than a translation exercise.
Built Into Every Department, Not Just the Front Desk
Once the front desk pilot proved successful, the same workflow was introduced across other guest-facing teams within three months.
That same pattern showed up in housekeeping. A guest called down about a broken air conditioner, speaking Mandarin. The housekeeper on duty spoke only Portuguese and took the request as easily as if the guest were speaking her own language. A technician was up within the hour, no callback to the front desk, no colleague pulled away to translate.
Department | What Changed |
Front desk | Check-in dropped from an average of 4 minutes to under 90 seconds for non-English-speaking guests |
Concierge | Guests could ask for restaurant or transport recommendations directly, without waiting for a translator |
Housekeeping | Room requests and maintenance issues went from averaging 2–3 conversations to resolving on the first one |
Restaurant & dining | Staff could explain dietary options and take special requests without guesswork |
Guest services | General requests throughout a stay were handled by whoever was on shift, not just multilingual staff |
Urgent situations | Medical requests and travel disruptions got handled with the clarity those moments require |
Every department used the same workflow. Staff didn't need to learn a different tool for each situation.
Why Self-Hosted Deployment Mattered Here
Guest conversations touch on payment details, passport information, and personal requests. That made deployment a real decision, not an afterthought.
The hotel chose a self-hosted setup. Speech recognition, translation, and voice synthesis all run on the hotel's own infrastructure. None of it passes through a third-party cloud service.
PolyTalk's self-hosted stack uses three components: faster-whisper for speech-to-text, an Ollama-compatible model for translation, and Piper for text-to-speech, all running inside the hotel's own environment.
For a property handling sensitive guest data across dozens of languages a week, keeping that data in-house wasn't a nice-to-have. It was the deciding factor between a vendor they could adopt and one they couldn't.
What Changed, in Practical Terms
Within three months of the first front-desk deployment, the differences were visible day to day rather than dramatic:
Front desk staff no longer needed to interrupt nearby colleagues to help with multilingual conversations.
Housekeeping and maintenance requests were resolved in one conversation instead of two or three.
Guests who used to type into a translation app started speaking normally again, and staff noticed conversations felt personal again, not transactional.
Coverage stopped depending on who was on shift. Any employee could handle any guest conversation.
As the front desk manager summed it up:
"We didn't hire for languages. We hired for hospitality. This just let that show up the way it was supposed to."
Choosing Real-Time Speech Translation Software for Hotels: What to Look For
Hotel managers evaluating real-time speech translation should consider four practical questions before choosing a platform:
Does it handle live conversation, or just messages?
Guest messaging and speech translation solve different problems. You likely need both.Does it fit your current workflow, or create a new one?
Staff shouldn't need to learn a separate process for international guests.Where does the audio go?
If guest conversations include payment or ID details, cloud-only processing may not meet your privacy requirements. Self-hosted or hybrid deployment gives you a say in that.Does it scale across departments without retraining everyone?
A tool that only works at the front desk solves a fraction of the problem.
PolyTalk was built around these four priorities: real-time speech translation, minimal workflow disruption, self-hosted deployment, and one platform that works across every guest-facing department. It's the same reasoning behind our broader hospitality use-case overview.
Curious how this workflow could fit your property?
See how PolyTalk enables real-time speech translation → without changing the way your staff already work.
FAQs
At this Lisbon property, check-in time for non-English-speaking guests dropped from an average of 4 minutes to under 90 seconds once front desk staff had PolyTalk in place.
Guest messaging is built for written, planned communication, confirmations, reminders, and updates. Real-time speech translation is built for live, face-to-face spoken conversation, which is where this hotel's biggest bottleneck was.
Yes. PolyTalk supports self-hosted deployment, so speech and translation processing can stay on infrastructure the hotel controls, rather than passing guest audio through a third-party cloud service.
Housekeeping and front desk saw the fastest, most measurable improvement, housekeeping requests went from averaging two to three conversations down to one. Concierge and dining followed once staff were used to the workflow.
Look for live conversation support (not just written translation), a workflow that fits existing operations, deployment flexibility for sensitive guest data, and coverage across every guest-facing department, not just the front desk.
Not exactly, the two solve different problems. PolyTalk doesn't replace your booking engine or PMS, it runs alongside it, handling the live conversations that happen once an international guest is in front of your staff, from check-in through checkout.
The most effective approach pairs real-time speech translation for live conversations with existing guest messaging tools for written updates. Speech translation lets staff and guests speak in their own languages during check-in, concierge requests, and service calls, without adding multilingual headcount.