Persona

This repository provides the PersonaPipeline for OpenVoiceOS (OVOS), which facilitates managing multiple personas and enables interactive conversations with a virtual assistant. The system is built around the concept of personas, each equipped with solvers to handle specific types of queries. This service allows you to load, register, and interact with personas using a chatbot API.

Features

• Multiple Personas: Manage a list of personas, each defined by a set of question-solving plugins.
• Dynamic: Switch personas on demand
• Conversational: Summon/Release personas to handle utterances instead of triggering intents
• Personalize: Create your own persona via .json files

Installation

pip install ovos-persona

Configuring Personas

Personas are loaded from configuration files, which can either be directly provided by plugins or user-defined JSON files.

By default, personas are loaded from the XDG configuration directory, just create .json files under ~/.config/ovos_persona

Personas work via solver plugins, each plugin is tried by order until one provides an answer

Find solver plugins here, some solvers may also use other solvers internally, such as a MOS (Mixture Of Solvers)

some repos and skills also provide solvers, such as ovos-classifiers (wordnet), skill-ddg, skill-wikipedia and skill-wolfie

Example to use a local OpenAI compatible server, ~/.config/ovos_persona/llm.json

{
  "name": "My Local LLM",
  "solvers": [
    "ovos-solver-openai-persona-plugin"
  ],
  "ovos-solver-openai-persona-plugin": {
    "api_url": "https://llama.smartgic.io/v1",
    "key": "sk-xxxx",
    "persona": "helpful, creative, clever, and very friendly."
  }
}

personas don't need to use LLMs, you don't necessarily need a beefy GPU to use ovos-persona

{
  "name": "OldSchoolBot",
  "solvers": [
    "ovos-solver-wikipedia-plugin",
    "ovos-solver-ddg-plugin",
    "ovos-solver-plugin-wolfram-alpha",
    "ovos-solver-wordnet-plugin",
    "ovos-solver-rivescript-plugin",
    "ovos-solver-failure-plugin"
  ],
  "ovos-solver-plugin-wolfram-alpha": {"appid": "Y7353-xxxxxx"}
}

this persona would:

• search online knowledge bases (wolfram alpha, wikipedia...)
• fall back to wordnet when offline for dictionary look up
• finally use a local chatbot (rivescript) for general chitchat
• the failure solver ensures the persona speaks an error if all solvers fail

Pipeline Usage

When configuring your pipeline it is suggested that "ovos-persona-pipeline-plugin-high" comes just before "fallback_high" and that "ovos-persona-pipeline-plugin-low" comes just before "fallback_low"

NOT YET FUNCTIONAL TODO: pending PR: OpenVoiceOS/ovos-core#570

The mycroft.conf configuration file can specify:

• The path to persona files (personas_path).
• A list of blacklisted personas (persona_blacklist) to not load.
• The default persona (default_persona).

{
  "intents": {
      "pipeline": [
        "...",
        "adapt_high",
        "...",
        "ovos-persona-pipeline-plugin-high",
        "...",
        "padatious_medium",
        "...",
        "ovos-persona-pipeline-plugin-low",
        "fallback_low"
    ],
    "ovos-persona-pipeline-plugin": {
      "personas_path": "/path/to/personas",
      "persona_blacklist": ["persona_to_exclude"],
      "default_persona": "default_persona"
    }
  }
}

NOTE: there is no "ovos-persona-pipeline-plugin-medium"

Direct Usage

from ovos_persona import PersonaService

# Initialize the PersonaService
persona_service = PersonaService(config={"personas_path": "/path/to/personas"})

# List all loaded personas
print(persona_service.personas)

# Ask a question to a persona
response = persona_service.chatbox_ask("What is the speed of light?", persona="my_persona")
print(response)

Each Persona has a name and configuration, and it uses a set of solvers to handle questions. You can interact with a persona by sending a list of messages to the chat() method.

from ovos_persona import Persona

# Create a persona instance
persona = Persona(name="my_persona", config={"solvers": ["my_solver_plugin"]})

# Ask a question to the persona
response = persona.chat(messages=[{"role": "user", "content": "What is the capital of France?"}])
print(response)

Events

• persona:query: Sent to submit a query to a persona.
• persona:summon: Sent when a persona is summoned.
• persona:release: Sent when a persona is released.

Contributing

Feel free to submit issues or pull requests for any improvements or bug fixes.