LIMMA

Language Interface Model for Machine Automation

LIMMA is a Python SDK that lets you control ESP8266/ESP32 devices using natural language commands.
It connects with the LIMMA API to translate user instructions into device function calls, then executes them over your local network.

---

Author: Yash Kumar Firoziya

---

✨ Features

Core Features

• 🔌 ESP Device Management – connect, reset, WiFi setup, status check
• 🌐 Natural Language → Device Control – powered by the LIMMA server
• 🧠 Context Manager – remembers previous commands for context-aware parsing
• 📡 Network Utilities – auto-discover ESP devices in your LAN
• ⚙️ Flexible Execution – supports wait(), device mapping, and reply messages

🆕 New in v0.2.0+ – Multi-Provider LLM Integration

• 🤖 Unified LLM Interface – Single API for multiple LLM providers via limma.llm
• 🔄 Provider Agnostic – Switch between OpenAI, Gemini, Mistral, Groq with one line
• 🧠 Consistent Message Format – Same request style across all providers
• 🔐 Flexible Configuration – Use env vars, inline setup, or config files
• 🚀 No Vendor Lock-In – Swap providers without rewriting logic
• 📜 Role-Based Conversations – System, user, assistant message support
• 🛡️ Automatic Token Handling – Prevents context overflows

🆕 New in v0.2.0+ – Voice Capabilities

• 🎙️ Cross-Platform Voice – Speech recognition & text-to-speech via limma.voice
• 🔊 Voice Customization – Adjust rate, volume, and gender (male/female/neutral)
• 🗣️ Simple API – Both standalone functions and VoiceAssistant class
• 🎯 Beginner-Friendly – Add voice control to your projects in minutes
• 🎚️ Voice Settings – List available voices, change gender on the fly

---

📦 Installation

pip install limma

Install with Optional Dependencies

# For LLM features
pip install limma[llm]

# For voice features
pip install limma[voice]

# For all features
pip install limma[all]

---

🚀 Quick Start

Basic ESP Control

from limma import Limma, LimmaConfig

config = LimmaConfig(
    esp_ip="192.168.1.100",
    application_type="home",
    device_map={"fan": "ch02", "light": "ch01"},
    api_key="your-api-key"
)

limma = Limma(config)
limma.execute_command("turn on the fan")

---

🆕 New in v0.2.0: LLM Integration

Unified Interface for Multiple LLM Providers

from limma.llm import config, generate, chat

# Configure once, use anywhere
config(
    provider="openai",  # or "gemini", "mistral", "groq"
    api_key="your-api-key",
    model="gpt-4"
)

# Generate text
response = generate("Explain IoT in simple terms")
print(response)

# Interactive chat
while True:
    user_input = input("You: ")
    if user_input.lower() == "exit":
        break
    print(f"AI: {chat(user_input)}")

Switch Providers Instantly

from limma.llm import config, generate

# Start with OpenAI
config(provider="openai", api_key="sk-...", model="gpt-4")
print(generate("Hello!"))

# Switch to Gemini (free tier available)
config(provider="gemini", api_key="AIza...", model="gemini-2.5-flash")
print(generate("Hello again!"))

# Switch to Groq for ultra-fast inference
config(provider="groq", api_key="gsk_...", model="mixtral-8x7b-32768")
print(generate("Fast response!"))

Using Environment Variables

# .env file
LLM_PROVIDER=openai
LLM_API_KEY=sk-your-key
LLM_MODEL=gpt-4

from limma.llm import config, generate

config()  # Auto-loads from environment
print(generate("What's new in Python 3.12?"))

Supported LLM Providers

Provider	Models	Use Case
OpenAI	GPT-4, GPT-3.5-turbo	General purpose, best quality
Google Gemini	Gemini 2.5/1.5 Flash	Free tier available, fast
Mistral	Mistral Large/Small	Open source, efficient
Groq	Mixtral, Llama 2	Ultra-fast inference

---

🆕 New in v0.2.0: Voice Integration

Simple Voice Control

from limma.voice import speak, listen

# Text-to-speech
speak("Hello! I'm your voice-enabled LIMMA assistant")

# Speech recognition
try:
    command = listen()
    print(f"You said: {command}")
    speak(f"Executing: {command}")
except Exception as e:
    speak("Sorry, I didn't catch that")

VoiceAssistant Class

from limma.voice import VoiceAssistant

# Create a customized voice assistant
assistant = VoiceAssistant(
    voice_rate=160,      # Words per minute
    voice_volume=0.8,    # Volume (0.0 - 1.0)
    voice_gender="female"  # male, female, or neutral
)

# Speak with the configured voice
assistant.speak("How can I help you with your ESP devices?")

# Listen for commands
command = assistant.listen()
if command:
    print(f"Command received: {command}")
    
# Change voice settings on the fly
assistant.set_voice_gender("male")
assistant.set_voice_rate(180)
assistant.speak("Voice settings updated")

Voice + ESP Control Combined

from limma import Limma, LimmaConfig
from limma.voice import VoiceAssistant

# Initialize voice and ESP control
voice = VoiceAssistant(voice_gender="female")
limma = Limma(LimmaConfig(esp_ip="auto", api_key="your-key"))

# Voice-controlled home automation
voice.speak("Voice control ready. Say a command.")

while True:
    command = voice.listen(timeout=5)
    if command:
        if "exit" in command.lower():
            voice.speak("Goodbye!")
            break
        success = limma.execute_command(command)
        if success:
            voice.speak("Command executed successfully")
        else:
            voice.speak("Failed to execute command")

Voice Customization Examples

from limma.voice import VoiceAssistant

va = VoiceAssistant()

# List available system voices
available_voices = va.get_available_voices()
print(f"Available voices: {available_voices}")

# Test different voice genders
va.set_voice_gender("female")
va.speak("This is the female voice")

va.set_voice_gender("male") 
va.speak("This is the male voice")

# Adjust speech rate and volume
va.set_voice_rate(200)  # Faster speech
va.set_voice_volume(0.5)  # Quieter
va.speak("This is fast and quiet")

va.set_voice_rate(120)  # Slower speech
va.set_voice_volume(1.0)  # Louder
va.speak("This is slow and loud")

---

🎯 Complete Example: Voice + LLM + ESP Control

from limma import Limma, LimmaConfig
from limma.voice import VoiceAssistant
from limma.llm import config as llm_config, generate

# Configure LLM for command understanding
llm_config(
    provider="gemini",  # Free tier
    api_key="your-gemini-key",
    model="gemini-2.5-flash"
)

# Initialize voice assistant
voice = VoiceAssistant(voice_gender="female")

# Setup ESP control
limma = Limma(LimmaConfig(
    esp_ip="auto",
    application_type="home",
    device_map={"fan": "ch02", "light": "ch01", "ac": "ch03"},
    api_key="limma-api-key"
))

# Intelligent voice-controlled automation
voice.speak("Smart home system activated")

while True:
    command = voice.listen()
    
    if command:
        if "exit" in command.lower():
            voice.speak("Shutting down")
            break
            
        # Use LLM to understand complex commands
        enhanced_command = generate(
            f"Convert this home automation request into a simple command: '{command}'. "
            f"Available devices: fan, light, ac. Response should be brief."
        )
        
        print(f"Original: {command}")
        print(f"Interpreted: {enhanced_command}")
        
        # Execute on ESP
        if limma.execute_command(enhanced_command):
            voice.speak("Done")
        else:
            voice.speak("I couldn't do that")

---

📚 API Reference

Core Modules

• Limma – Main SDK controller
• LimmaConfig – Configuration container
• ESPManager – ESP device operations
• ContextManager – Command history and context
• NetworkUtils – Network discovery utilities

🆕 LLM Module (limma.llm)

• config(**kwargs) – Set provider, API key, model
• generate(prompt, **kwargs) – Single text generation
• generate_stream(prompt, **kwargs) – Streaming response
• chat(message) – Conversational interface
• reset_chat() – Clear conversation history

🆕 Voice Module (limma.voice)

• speak(text, **kwargs) – Text-to-speech conversion
• listen(timeout) – Speech recognition
• VoiceAssistant – Class-based voice interface
  • set_voice_rate(rate) – Adjust speech speed
  • set_voice_volume(volume) – Adjust volume
  • set_voice_gender(gender) – Change voice gender
  • get_available_voices() – List system voices
  • simple_conversation(prompt) – Quick Q&A

---

🛡️ Error Handling

from limma.voice.exceptions import ListenTimeoutError, AudioCaptureError
from limma.llm.exceptions import AuthenticationError, ModelNotFoundError

try:
    response = generate("Hello")
except AuthenticationError:
    print("Check your API key")
except ModelNotFoundError:
    print("Invalid model name")

---

📦 Dependencies

• Core: requests
• LLM Module: requests (no additional deps)
• Voice Module: SpeechRecognition, pyttsx3, pyaudio

---

📄 License

Licensed under the Apache License 2.0. See LICENSE for details.

---

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

---

⭐ Support

If you find LIMMA useful, please give it a star on GitHub!

---

Now LIMMA is not just an ESP control SDK – it's a complete toolkit for building intelligent, voice-controlled, multi-provider AI automation systems. 🚀