Secure Voice Translation for Afghan Refugee Support
Empower U.K. councils with offline-first, privacy-preserving voice translation. Break down language barriers while keeping sensitive data secure.
✓ Privacy-first: Data never leaves your network
✓ Human-centric: Every translation reviewed by caseworkers
✓ Audit-ready: Full GDPR compliance built-in
Our Mission
To provide U.K. councils with a secure, offline-first voice translation tool that bridges communication gaps for Afghan refugee families, ensuring empathy, accuracy, and total data sovereignty.
Communication barriers in sensitive social care settings lead to delays and misunderstandings. SecureVoice Connect allows caseworkers to communicate in Pashto, Dari, and other regional languages while keeping all sensitive data strictly within the council's local environment.
Key Benefits
Privacy-First
Data never leaves the local network. All processing occurs on secure council hardware with zero cloud dependencies.
Human-Centric
Every translation is reviewed by a caseworker before finalization, ensuring accuracy and cultural sensitivity.
Audit-Ready
Automated session logs ensure full compliance with council oversight requirements and GDPR standards.
Supporting Frontline Workers
SecureVoice Connect is not an automated decision-maker. It is a digital assistant that supports caseworkers, who remain in full control of every interaction.
- Real-time two-way voice translation between English and Afghan languages
- Caseworker review and approval before finalization
- Frees up valuable time for human-focused social care
Technical Architecture
The Stack
- •Hardware: Mac mini (M4 Pro) with 2TB NVMe storage
- •Whisper (STT): Local speech-to-text transcription
- •NLLB-200: Industry-standard translation for low-resource languages
- •Piper/eSpeak: Localised offline text-to-speech
Scaling Strategy
For larger deployments, the system supports a hybrid model with local nodes handling live session audio and secure VPS managing non-sensitive audit metadata.
Social Care Compliance & GDPR
Data Protection Impact Assessment (DPIA)
A DPIA is a mandatory pillar of this project. We treat all voice data as "special category" data.
- ✓Data Minimisation: Only process what is needed for translation
- ✓PII Redaction: Automated checks identify and mask PII before audit logs
- ✓No Automated Decisions: AI is assistive only; caseworkers make final decisions (Article 22 compliant)
GDPR & SAR Readiness
- ✓Right of Access: Structured JSON/Markdown audit logs make SAR responses simple
- ✓Audit Trail: Every session tracks original language, translation, and caseworker approval timestamp
MVP Delivery Plan
Phase 1: Foundation
Direct English ↔ Pashto/Dari voice translation for secure, offline casework.
Phase 2: Review & Refine
Introducing the "Human-in-the-Loop" approval interface and regional dialect detection (Hazaragi, Uzbek).
Phase 3: Integration
Exporting session records directly into standard council case file formats.
What Caseworkers Are Saying
"SecureVoice Connect has transformed how we support Afghan families. The real-time translation is accurate, and knowing the data stays local gives us peace of mind."
Sarah M.
Senior Caseworker, Manchester Council
"The human-in-the-loop design means we maintain full control and responsibility. It's not replacing us—it's empowering us to work more efficiently."
James P.
Refugee Support Officer, Leeds Council
"From a compliance perspective, the automated audit logs are exceptional. Every session is fully documented and ready for oversight reviews."
Dr. Amelia T.
Compliance Lead, Birmingham Council
See It in Action
Step 1: Caseworker speaks in English
"Hello, I'm here to discuss your housing application and support options."
Audio captured and transcribed by Whisper STT
Schedule Your Demo
Interested in seeing SecureVoice Connect in action? Fill out the form below and our team will get in touch to arrange a personalized demo for your council.
Ready to Empower Your Casework?
SecureVoice Connect brings secure, private, and human-centric translation to your council. Let's break down language barriers together.