Node.js AI Voice Translation Center

PROJECT DESCRIPTION
Context
I'm looking for an experienced developer to build an intelligent call center system with automatic real-time voice translation. The project aims to enable agents in India (English speakers) to receive and handle calls from French customers, with instant bidirectional translation.
Objective
Create a complete solution that enables:

Receiving incoming calls on a French number (OVH Telecom)
Real-time voice translation French ↔ English/Hindi
Automatic transfer to agent
Latency < 1 second
GDPR compliant (EU region)


REQUIRED TECH STACK
Backend

Node.js (v20+)
Express.js for HTTP server
WebSocket (ws) for real-time audio streaming
PM2 for process management

APIs & Services

OVH Telecom: VoIP telephony (REST API)
Deepgram Nova-2: Speech-to-Text (French & English)
DeepL API: Translation (FR↔EN, FR↔HI)
Google Cloud Text-to-Speech Neural2: Voice synthesis (europe-west1 region)

Infrastructure

OVH VPS with Plesk Obsidian (Ubuntu 24)
Nginx reverse proxy (configuration provided)
SSL/HTTPS Let's Encrypt (already configured)
Subdomain: callcenter.nexus-sirh.fr


REQUIRED FEATURES
1. Incoming call management

Webhook to receive OVH calls
Customizable greeting message (voice synthesis)
Automatic transfer to agent number
Error handling and timeouts

2. Real-time translation pipeline
FR Audio → Deepgram STT → DeepL → Google TTS EN → Agent
EN Audio → Deepgram STT → DeepL → Google TTS FR → Client
3. Performance optimizations

Smart cache for frequent translations
Optimized audio buffering (32-frame chunks)
Total latency < 1000ms
Efficient memory management

4. Web dashboard

Homepage with real-time statistics
Active calls counter
Average latency
System health monitoring (/health endpoint)

5. Logs & Monitoring

Detailed logs (PM2)
Per-call metrics (duration, latency, errors)
Cache statistics


DELIVERABLES
Source code

callcenter-server.js: Complete main server
package.json: Dependencies
.env.example: Configuration template
test-translation.js: API testing script
Clean, commented code, ES6+

Documentation

README.md: Installation, configuration, usage
DEPLOYMENT.md: Plesk/VPS deployment guide
API.md: Endpoints documentation
Architecture diagram

Configuration

Nginx configuration files (provided by me)
PM2 configuration for auto-start
Deployment scripts

Testing & Validation

End-to-end functional tests
Latency validation < 1000ms
Testing with real OVH calls


REQUIRED SKILLS
Essential

Node.js / Express.js (expert level)
WebSocket real-time
REST APIs (third-party integration)
Audio streaming (buffers, encoding)
VoIP / Telephony (OVH or Twilio)

Important

Google Cloud APIs (TTS)
Deepgram / STT APIs
DeepL API
PM2 / Process management
Nginx (reverse proxy)

Bonus

Call center / cloud telephony experience
Real-time performance optimization
GDPR / Sensitive data handling
Docker (optional)


PROVIDED TECHNICAL CONTEXT
Existing infrastructure

OVH VPS (8GB RAM, 4 vCPU)
Plesk Obsidian v18.0.73
Node.js v20.20.0 installed
Nginx configured
Let's Encrypt SSL active
Domain: callcenter.nexus-sirh.fr

API accounts (provided)

OVH Telecom (French number)
Deepgram ($200 free credits)
DeepL (500k characters/month)
Google Cloud (configured project, EU region)

Access provided

SSH / Plesk web terminal
FTP/File manager access
API keys (privately)


BUDGET & TIMELINE
Budget range

$500 - $1,300 depending on experience and portfolio
Payment milestones:

30% upon detailed specification approval
40% upon functional code delivery
30% after testing and production deployment



Timeline

Phase 1 (3-5 days): Architecture + Basic server code
Phase 2 (3-5 days): API integration + Translation pipeline
Phase 3 (2-3 days): Testing + Optimization + Documentation
Total: 8-13 days (negotiable)


SELECTION CRITERIA
Required in your proposal

Portfolio: Similar projects (Node.js + real-time APIs)
VoIP experience: Have you worked with OVH Telecom / Twilio?
Availability: How many hours/day can you dedicate?
Methodology: How do you plan to approach this project?
Questions: What additional information do you need?

What makes you stand out

Proven experience in cloud telephony
Open-source projects / Active GitHub
Performance optimization expertise
Clear and responsive communication
Availability for post-delivery support (optional)


HOW TO APPLY
Start your message with "OVH-AI-CALLCENTER" to show you've read the entire post.
Include:

Link to your portfolio/GitHub
Your VoIP/telephony experience
Your estimated delivery timeline
Your rate (fixed price or hourly)
One technical question about the project


IMPORTANT

No Node.js beginners
No agencies (individual freelancer only)
No generic templates / AI responses
Original code with transferred ownership
Communication in English or French
Availability for 2-3 video calls during project


ADDITIONAL INFORMATION
Real use case
A French customer calls the OVH number → Hears "Hello, you'll be connected" → Agent answers → Everything the customer says in French is translated to English for the agent → Everything the agent says in English is translated to French for the customer → Imperceptible latency.
Possible future enhancements

Call recording
Advanced analytics dashboard
Multi-language support (Spanish, Arabic, etc.)
CRM integration


Ready to create an innovative solution to break language barriers?
Apply now!

Project type: Fixed-price project
Duration: 1-2 weeks
Level: Expert
Languages: English, French