AI Translator Profiles Module Documentation

Module Overview (Technical)
Module Overview (Commercial/Business)
Module Overview (End User/Administrator)
Configuration Sections
Settings Reference
Common Scenarios & Examples
Limitations & Important Notes
Troubleshooting Tips

---

1. Module Overview (Technical)

What Are AI Translator Profiles?

AI Translator Profiles replace legacy static language fields with a dynamic, relational, UUID-based architecture. Instead of hardcoding a "Native Language" directly on an extension, administrators assign an AI Translator Profile to the extension. This profile governs exactly how real-time simultaneous translation behaves for that user, including the source/target language, AI provider, TTS (Text-to-Speech) neural voice settings, and translation strictness.

Architecture

Plaintext

┌─────────────────────────────────────────────────────────────────┐
│              AI Translator Architecture                        │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  AI Translator Profile: "Executive Spanish Translation"        │
│  ┌──────────────────────────────────────────────────────────┐   │
│  │  Profile ID: [UUID]                                      │   │
│  │                                                          │   │
│  │  Translation Settings:                                   │   │
│  │  ├─ User's Native Language: Spanish                      │   │
│  │  ├─ Target Business Language: English                    │   │
│  │  └─ Translation Mode: Simultaneous (Streaming)           │   │
│  │                                                          │   │
│  │  TTS Voice Engine (Azure Neural Resonance):              │   │
│  │  ├─ Voice: es-MX-JorgeNeural                             │   │
│  │  ├─ Style: Professional                                  │   │
│  │  └─ Speed: 1.1x                                          │   │
│  │                                                          │   │
│  │  AI Model:                                               │   │
│  │  ├─ Provider: OpenAI                                     │   │
│  │  └─ Model: GPT-4o Realtime                               │   │
│  └──────────────────────────────────────────────────────────┘   │
│       ▲                                                         │
│       │ Linked via aiTranslatorProfileId                        │
│  ┌────┴─────────────────────────────────────────────────────┐   │
│  │  Extension 1001 (CEO)                                    │   │
│  │  - simultaneousTranslationEnabled: true                  │   │
│  │  - aiTranslatorProfileId: [UUID]                         │   │
│  └──────────────────────────────────────────────────────────┘   │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

Key Tables

Table	Purpose
`public.ai_translator_profiles`	Core configuration for the translation profiles (UUID-based).
`public.sip_extensions`	Contains `ai_translator_profile_id` linking users to the translation configuration.

---

2. Module Overview (Commercial/Business)

Business Value

The AI Translator Profiles module breaks down language barriers in enterprise environments. It enables real-time, simultaneous translation during live phone calls.

Legacy System	New AI Translator Architecture
Basic translation requiring human interpreters	Real-time AI interpretation with ultra-low latency
Hardcoded per user	Centralized profiles that can be updated in bulk
Monotone robotic voices	Azure Neural Resonance for human-like inflection

Use Cases

Multinational Internal Communications

- An engineer in Japan calls a manager in the US. Both speak in their native languages; the PBX translates the audio in real-time.

Global Customer Support

- Support agents can assist callers in any language without needing to learn the language. The profile ensures the customer hears a highly professional, native-sounding TTS voice.

---

3. Module Overview (End User/Administrator)

What Can You Do?

Create centralized translation profiles instead of configuring languages one-by-one per user.
Configure exact Voice profiles (TTS) that represent the translated speech.
Easily manage configurations via a standardized 4-column DataGrid.

Go to Admin → AI → Translator Profiles.
Click + Add to create a new profile.
Configure the source/target languages and select the AI model.
Go to PBX → Extensions, edit an extension, and in the Advanced tab, assign the profile.

---

4. Configuration Sections

General Information

Field	Description	Example
Profile Name	A descriptive name for the configuration.	`Sales - English to Spanish`
Description	Internal notes.	`Used for LATAM sales reps`

Language & Translation Settings

Field	Description	Example
Native Language	The language spoken by the extension owner.	`Spanish (es-MX)`
Target Language	The default language the AI should translate into.	`English (en-US)`
Translation Engine	The LLM provider performing the translation.	`OpenAI`
Model	Specific model (e.g., GPT-4o) optimized for speed.	`GPT-4o`

Voice Settings (TTS)

Field	Description	Example
Voice Provider	The engine used to synthesize the translated text back into audio.	`Azure Neural Resonance`
Voice Selection	The specific voice persona.	`en-US-JennyNeural`

---

5. Settings Reference

TIP

Voice Selection: For simultaneous translation, it is highly recommended to use Azure Neural Resonance voices, as they provide the lowest latency and highest natural inflection for real-time streaming.

WARNING

Latency Expectations: Simultaneous translation requires processing time. Expect a 1.5 to 2.5-second delay between speech and translation depending on the chosen model's context-window processing.

---

6. Common Scenarios & Examples

Scenario: Creating a Profile for a French Executive

Navigate to AI Translator Profiles and click + Add.
Name: French Exec Translation
Native Language: French
Translation Engine: OpenAI (Model: GPT-4o)
Voice Provider: Azure Neural
Voice Selection: fr-FR-HenriNeural
Save the profile.
Navigate to Extensions, edit the executive's extension, go to the Advanced tab.
Toggle Simultaneous Translation to Enabled.
Select French Exec Translation from the dropdown.

---

7. Limitations & Important Notes

UUID Architecture: Translator profiles use UUIDs, meaning they are globally unique and immune to ID collisions during database merges or migrations.
Action Bar Parity: The module supports standard DataGrid actions including Edit, Duplicate, and Delete directly from the fixed Action Bar.

---

8. Troubleshooting Tips

Symptom	Possible Cause	Solution
Translation is heavily delayed	Wrong model selected	Ensure you are using `GPT-4o` or models specifically tagged for low-latency streaming.
Robotic voice	Using legacy TTS	Switch Voice Provider to Azure Neural Resonance.
Profile doesn't apply to calls	Not assigned to extension	Verify the profile is selected in the Extension's Advanced tab and the feature is toggled On.

---

Glossary

Term	Definition
Translator Profile	Settings mapping source/target languages and voices for translation
Simultaneous Translation	Real-time audio translation during a call

---

Lua Script Status ℹ️

Integration Module

Component	Status	Notes
AI Translator Profiles	ℹ️ Configuration module	Translator parameters loaded dynamically by translation Lua handler

--- Documentation last updated: May 2026