Skip to content

Performing a RAG (Retrieval Augmented Generation) assessment using voice-to-voice query resolution. Provide the file containing the queries, ask the questions, and receive the results via voice.

Notifications You must be signed in to change notification settings

Adii2202/RAG-AI-Voice-assistant-

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RAG AI Voice Assistant

Block Diagram

Overview

The RAG AI Voice Assistant is a comprehensive system designed to interact with users through spoken commands and questions. It leverages various components to provide a seamless user experience:

  • User: Interacts with the voice assistant by providing spoken instructions or questions.
  • Voice to Text Conversion: Converts the user's speech into text for computer processing.
  • Faster Whisper Model: Recognizes speech using the Faster Whisper model.
  • OLLAMA & MISTRAL: Large language models (LLMs) used for processing user queries and generating responses.
  • RAG Retrieval Augmented Generation: Combines information from conversation history and the internet to generate responses.
  • Chat Memory: Stores past conversations with the user.
  • Text to Voice Conversion: Converts generated text responses back into speech.
  • Voice Assistant Bot: Integrates all components to enable voice-based interaction.

Components

Memory Buffer

Provides fast retrieval of data for quick access.

# Example code for memory buffer usage
memory_buffer = MemoryBuffer()
data = memory_buffer.retrieve_data(key)

Huge Data Storage

Utilizes a vector database for efficient storage and retrieval of large datasets.

# Example code for vector database usage
vector_db = VectorDatabase()
vector_db.store_vectors(vectors)
results = vector_db.search(query_vector)

Models

Faster Whisper

An implementation of a model for speech recognition.

# Example code for Faster Whisper model usage
faster_whisper = FasterWhisperModel()
transcription = faster_whisper.transcribe(audio)

Mistral

An implementation of a model for natural language understanding and generation.

# Example code for Mistral model usage
mistral = MistralModel()
response = mistral.generate_response(query)

Qdrant

Qdrant is a vector search engine designed for similarity search, nearest neighbor search, and clustering of high-dimensional data.

Docker Installation

To use Qdrant, pull the Docker image and run it as a container:

docker pull qdrant/qdrant
docker run -p 6333:6333 qdrant/qdrant

For detailed instructions, visit the Qdrant website.

Usage

  1. Memory Buffer: Utilize the memory buffer for fast data retrieval.
  2. Huge Data Storage: Store and retrieve large datasets efficiently using the vector database.
  3. Models: Use Faster Whisper for speech recognition and Mistral for natural language processing.

Contributing

Contributions are welcome! Feel free to submit pull requests or open issues for any improvements or suggestions.

About

Performing a RAG (Retrieval Augmented Generation) assessment using voice-to-voice query resolution. Provide the file containing the queries, ask the questions, and receive the results via voice.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages