Closing the Infrastructure Gap in Institutional Multilingualism

Large institutions can move from legacy remote interpreting models to a permanent multilingual voice infrastructure by shifting their focus from external, service-based procurement to an integrated technology layer. Rather than treating translation as a siloed task performed by external vendors, modern organizations are beginning to view multilingualism as an architectural requirement. This transition involves embedding real-time speech-to-speech capabilities directly into existing communication tools, ensuring that language access is a continuous, native feature of the enterprise environment rather than an intermittent external service.

Recent procurement activities highlight the scale of the challenge. The city of Boston, for example, is currently seeking bids for a $3 million contract to provide 24/7 Over-the-Phone Interpreting (OPI) and Video Remote Interpreting (VRI) across 17 languages. Simultaneously, the European Central Bank has opened tenders for translation services from English into multiple European languages. These high-stakes requirements demonstrate a massive, growing need for reliable multilingual communication, yet they also reveal a reliance on traditional, service-based models that often remain disconnected from the core operational workflows of the institution.

The Limitations of the Service-Based Model

Traditional OPI and VRI models were designed for a different era of communication. In these models, when a language barrier arises, a participant must step out of their primary workflow to connect with an external interpreter. This creates several friction points that hinder modern institutional efficiency. First is the issue of context; an external interpreter often lacks the specific organizational knowledge required for complex technical or administrative discussions. Second is the technical fragmentation; the interpreting session usually happens on a third-party platform or via a separate phone line, breaking the continuity of the digital record and the user experience.

Furthermore, the "on-demand" nature of these services introduces latency. In critical environments—such as emergency services, central banking, or municipal governance—the seconds or minutes spent waiting for an interpreter to connect can have significant consequences. As institutions grow more distributed and globalized, the need for a more immediate, persistent solution becomes clear. The goal is no longer just to "find a translator" but to ensure that the communication infrastructure itself is inherently multilingual.

From Services to Multilingual Voice Infrastructure

A multilingual voice infrastructure represents an architectural shift. Instead of treating translation as a bolt-on service, it functions as a foundational layer that processes speech in real time across any application the organization already uses. This approach allows employees and citizens to speak their native languages while the infrastructure handles the translation and routing behind the scenes, preserving the original communication workflow.

Hitoo is being developed to provide this exact layer of infrastructure. By focusing on the endpoint and the communication tools that organizations already trust, Hitoo aims to eliminate the need for external silos. Hitoo Desktop, currently under development for macOS and Windows, is designed to connect microphones, speakers, and live communication software directly to a multilingual speech layer. This architecture is intended to allow organizations like the City of Boston or the ECB to maintain their existing workflows while adding a high-performance translation capability that is always available.

Enterprise Governance and Deployment

For large-scale institutions, moving to a technology-first infrastructure also addresses critical concerns regarding governance and privacy. Traditional OPI/VRI models often involve a rotating cast of human contractors, making centralized data policy and auditability difficult to manage at scale. An infrastructure-based approach allows for centralized management and deployment.

Hitoo Enterprise is being designed as a managed endpoint infrastructure specifically for these complex environments. The architecture is intended to support organization-wide user management, device enrollment, and centralized policies. This direction includes providing operational telemetry and administrative auditability, ensuring that every multilingual interaction complies with strict institutional data sovereignty and privacy requirements. By moving translation onto the internal infrastructure, organizations gain a level of control and security that external service bureaus cannot easily replicate.

The Research Frontier of Real-Time Speech

The future of institutional communication lies in reducing the technical and cognitive distance between speakers of different languages. Current cascaded architectures—which recognize speech, translate the text, and then synthesize a new voice—often introduce latency and strip away the nuances of the original speaker’s tone. Reducing this latency is essential for natural, full-duplex conversations where participants can interrupt or react in real time.

AURIS, Hitoo’s proprietary research initiative, is investigating a more direct speech-to-speech translation architecture. The research objective is to move away from cascaded models in favor of a direct path that preserves the speaker’s voice identity and expression while targeting end-to-end latency below 400 ms. While this is currently a research frontier, it points toward a future where multilingual voice infrastructure feels as natural and immediate as a monolingual conversation. For institutions managing high-stakes international relations or essential public services, this evolution from "interrupted service" to "integrated infrastructure" is the necessary next step.