Usage

How It Works

Call Initiation

The process begins when an incoming call is routed to the Virtual Translator infrastructure via SIP Trunk. The SIP Gateway acts as a bridge between the Contact Center Platform and the product, enabling voice streaming between all parties.

Direct Communication

Before translation is activated, the SIP Gateway streams audio directly between the caller and agent, allowing them to communicate naturally. If both parties understand each other, no translation is needed.

Activating Translation

When the agent encounters a language barrier, they can activate translation through the Agent Copilot interface.

Translation Pipeline

Once activated, the Orchestration Service processes the voice stream through the following stages:

1. Voice Activity Detection: Identifies when speech occurs, filtering out silence and background noise for efficient processing.

2. Language Detection: Automatically identifies the caller's spoken language.

3. Speech Recognition: Transcribes spoken words into text in real time.

4. Translation: Converts the transcribed text from the source language to the target language.

5. Text-to-Speech Synthesis: Generates natural-sounding audio from the translated text.

Delivering the Translation

The translated audio stream is sent back through the SIP Gateway to the agent. The same pipeline processes the agent's responses in reverse, delivering translated audio to the caller. This bidirectional process continues seamlessly throughout the call.

Agent Copilot

Throughout the call, the Agent Copilot provides a real-time user interface that lets agents view live transcriptions, translations, and sentiment indicators. Agents can also adjust audio playback preferences, manually override language settings, or deactivate translation, all without interrupting the conversation.