Back to Projects
Python

Hebrew-Call-Center-Agent

The agent simulates a Hebrew customer support call by converting typed messages into vowelized Hebrew speech, then transcribing the audio back to text and logging the full conversation. It automates, records, and validates interactions for efficient, high-quality multilingual support.

View Source Code

System Overview

What the project does

Simulates a Hebrew‑language customer‑support call for TV subscription cancellations using a multi‑agent CrewAI framework, generating both text transcripts and audio for each dialogue step.

Key features

  • - Hebrew text handling with optional nikud (vowel marks) via Phonikud ONNX model
  • - Text‑to‑Speech (Chatterbox‑tts) and Speech‑to‑Text (OpenAI Whisper) integration
  • - Structured multi‑agent conversation flow with guardrails (max 6 turns, token monitoring, error handling)
  • - Automatic logging of full transcript, audio files per step, and execution logs

    • Tech stack

  • - Python 3.10–3.11
  • - CrewAI for agent orchestration
  • - OpenAI Whisper API for STT, OpenAI LLM for dialogue generation
  • - Phonikud ONNX model for Hebrew nikud processing
  • - Chatterbox‑tts for Hebrew TTS
  • - YAML configuration for agents and tasks

    • Use case

    Demonstrates end‑to‑end Hebrew voice‑based customer service automation, useful for training AI agents, prototyping call‑center workflows, or creating multilingual support demos.

    Architecture Details

    This system integrates multiple components for a seamless automation flow. Structural interpretation based on project focus:

    Backend Infrastructure

    Core execution layer for robust data processing and API handling.

    AI / Logic Core

    Intelligent decisioning via models or logical workflow rules.

    Tech Stack

    PythonIntegrationAutomationAPIs

    Key Capabilities

    • ▹ Custom workflow execution
    • ▹ Data transformation and routing
    • ▹ Extensible architecture