Project Overview I need a cloud-based AI receptionist system for my medical office (psychiatrist practice in Greece). The system should answer incoming calls, interact with patients in Greek, record and transcribe the conversation, and automatically send me an email with the caller’s number, transcript, and audio recording. It should also be able to make outbound reminder calls (e.g., appointment confirmations) using text-to-speech (TTS). The system must be open-source based, self-hosted on a VPS, and not depend on proprietary SaaS. Preferably built using Asterisk / FreePBX / Jambonz / Vosk / Piper / Ollama or similar open technologies. Required Features 1. Inbound Calls Answer calls automatically with a Greek TTS greeting (e.g., “Καλησπέρα σας, καλέσατε το ιατρείο του Δρ. Καραΐσκου…”). Record the entire call (MP3 or WAV). Perform speech-to-text (STT) transcription (Greek language). Identify the caller’s request using a local AI model or rule-based logic (e.g., “appointment,” “prescription,” “information”). Reply with a short Greek text-to-speech (TTS) message. Send me an email notification after the call with: Caller number (CLI) Transcript (text) AI reply (if any) Link or attachment of the audio file 2. Outbound Calls Ability to call patients (from the clinic’s phone number) and play a pre-recorded or TTS message, e.g. “Your appointment with Dr. Karaiskos is confirmed for tomorrow at 10:00.” Optionally log or email call status (delivered / answered / failed). 3. Email Notification System SMTP integration (Gmail / custom domain) Template email with: Caller ID Transcript (text) Audio file (MP3 or link) Optional: timestamp, call duration, and AI classification (“appointment request,” “general inquiry,” etc.) Technical Requirements Component Technology (preferred) Telephony PBX Asterisk / FreePBX / Jambonz (open-source) Speech-to-Text (STT) Vosk (offline Greek model) or Google Speech API Text-to-Speech (TTS) Piper (offline Greek) or Google TTS AI Logic Ollama (local LLM) or simple Python intent parser Email Python SMTP script / n8n automation Hosting VPS (Hetzner, Contabo, or DigitalOcean – 2 vCPU / 4 GB RAM / 40 GB SSD) Operating System Ubuntu 22.04 or Debian 12 Functional Goals Fully working inbound/outbound call system accessible via web interface (FreePBX GUI or similar). Works with my existing phone line (Nova / Cosmote) via SIP trunk or FXO gateway. All voice and text data remain private (self-hosted → GDPR-compliant). Scalable for ~2000–4000 minutes/month. Simple interface or configuration file for updating greeting messages and email recipients. Deliverables 1. Cloud VPS setup (Ubuntu + Asterisk or equivalent) 2. Configured STT, TTS, and AI modules (Greek language) 3. Email notification automation (Python or n8n) 4. Web interface (FreePBX or similar) 5. Step-by-step documentation (login, settings, backup) 6. One live test call (inbound and outbound) confirming the system works end-to-end. Budget Range €250 – €400 one-time setup cost Cloud VPS hosting (~€5–10/month) Optional API costs (Google Speech/TTS ~€0.01/minute if used) Ongoing self-hosting with minimal or zero subscription fees Additional Notes System language: Greek (voice + transcription). No proprietary Twilio or Dialogflow dependencies unless open-source alternatives are unavailable. Must be easily portable to a local server later (same setup on-premise). Developer should deliver SSH access and full control of the VPS. Ideal Freelancer Experience with Asterisk / FreePBX / VoIP / SIP trunking Familiar with speech recognition (Whisper, Vosk) and TTS systems (Piper, Coqui, Google) Knowledge of Python / Bash scripting for automation Prior work on AI-assisted telephony or voice bots preferred