Configuration File: tts_engines.json

The tts_engines.json file, located in /system/tts_engines/, stores information about the Text-to-Speech (TTS) engines available in the application. It keeps track of the currently loaded engine, its selected model, and a list of other available engines with their respective models.

File Structure

{
    "engine_loaded": "piper",
    "selected_model": "piper",
    "engines_available": [
        {
            "name": "parler",
            "selected_model": "parler - parler_tts_mini_v0.1"
        },
        {
            "name": "piper",
            "selected_model": "piper"
        },
        {
            "name": "vits",
            "selected_model": "vits - tts_models--en--vctk--vits"
        },
        {
            "name": "xtts",
            "selected_model": "xtts - xttsv2_2.0.3"
        }
    ]
}

Main Settings

These settings will be what is used when you start AllTalk.

• engine_loaded: The name of the currently loaded TTS engine.
• selected_model: The model currently being used by the loaded engine.

Available Engines

The engines_available array contains objects for each available TTS engine:

• name: The name of the TTS engine.
• selected_model: The currently selected model for that engine.

Functionality

1. Engine Switching: When you switch to another TTS engine in the interface, the engine_loaded and selected_model fields are updated to reflect the newly selected engine and its corresponding model.

2. Model Switching: If you change the model for the currently loaded engine, only the selected_model field is updated for the engine_loaded.

3. Persistence: This file helps maintain the state of TTS engine selections across sessions, allowing the application to remember your last used engine and model and load that in when you next start AllTalk.

Example Scenarios

1. Current State:

   • Loaded engine: Piper
   • Selected model for Piper: "piper"

2. Switching to VITS:

   • engine_loaded would change to "vits"
   • selected_model would update to "vits - tts_models--en--vctk--vits"

3. Changing Piper's model (while Piper is the loaded engine):

   • Only the selected_model field would update to reflect the new Piper model

This configuration allows for flexible management of multiple TTS engines and their respective models, enabling easy switching and state management within the application.