SUMMARY
SentiVue lets people speak in their own language and be understood, in any voice conversation that matters.
When two people who speak different languages get on a phone call, SentiVue sits inside the call and translates between them as they speak. Each person speaks naturally in their own language and hears the other in theirs. No app to install. No hardware to fit. The same underlying capability also runs AI voice agents that answer calls and switch languages mid-conversation, and live event interpretation where audiences follow speakers in the language they understand.
The depth comes from building speech models for languages that global providers under-serve. Through a partnership with CNCA and BSC AI Factory, using Deucalion, Portugal’s national high-performance computing infrastructure, a European Portuguese voice model is already in production, with more languages on the way.
The result is direct. Organisations serve more people, in more languages, on the systems they already use.
PROBLEM
Voice communication still runs on infrastructure designed for one language at a time. Phone systems, contact centres, public services, and live events were built around the assumption that everyone in the conversation speaks the same language. That assumption no longer holds.
The cost is felt everywhere. Customers wait longer or get transferred until someone bilingual is found. Calls break down. Patients, residents, and travellers who do not speak the local language are quietly excluded from services they are entitled to.
Hiring multilingual teams does not scale. Interpreter services are slow, expensive, and awkward to fit into live calls. AI voice agents break the moment a caller switches language, and global cloud providers offer thin coverage for languages like European Portuguese.
The gap sits exactly where the conversation happens, and it has not been closed.
SOLUTION
SentiVue is building the voice layer that makes multilingual conversation work in real time.
The starting point is the call itself. When two people speak different languages on a phone call, SentiVue sits inside the call and translates between them as they speak. Each person speaks in their own language and hears the other in theirs. No app to install. No hardware to fit. Latency short enough that the conversation feels natural rather than mechanical.
The same capability extends naturally to the rest of voice. It can answer calls on a business's behalf, handle what the caller needs, switch languages mid-conversation, and transfer cleanly to a human operator with the translation still active. It can run on a stage, so a speaker reaches an audience listening in several languages at once.
What makes this credible is the model layer underneath. Real-time voice translation demands speech models tuned for each language, and global providers under-serve many of them. SentiVue is closing that gap by building its own. Through a partnership with CNCA and BSC AI Factory, using Deucalion, Portugal’s national high-performance computing infrastructure, a European Portuguese voice model is already in production, with more languages in development.
The impact is direct. Organisations serve customers across languages without rebuilding their phone systems or hiring large multilingual teams. Public services reach people in their own language. Events open up to wider audiences. Voice becomes multilingual by default, on the infrastructure that already exists.