AMEERAZAM08 / ML-QA

🦒 Colab

Colab	Info
	ML-QA Chatbot 🚀

Chatbot Interaction Demo

Overview

This Space demonstrates how to interact with a chatbot using open accessible models. It offers a seamless experience of converting speech to text, generating chatbot responses, and then converting these responses back to speech.

Models Used

Speech to Text

• Model: Whisper-large-v3
• Source: Whisper-large-v3 on Hugging Face
• Description: This ASR (Automatic Speech Recognition) model transcribes recorded audio to text. It is integrated through a Gradio client.

Chat Model

• Model: Mixtral-8x7B-Instruct-v0.1
• Source: Mixtral-8x7B-Instruct-v0.1 on Hugging Face
• Description: This Large Language Model (LLM) Mistral serves as the chat model, generating responses based on the transcribed text.

Text to Speech

• Model: Facebook Seamless T2ST
• Source: Facebook Seamless T2ST on GitHub
• Description: This Multilingual TTS (Text-to-Speech) model generates audio from the chatbot's text responses. The model is hosted locally.

Preview

Installation

Use the following commands to install the necessary SDK and dependencies:

Local installation:

git clone https://github.com/AMEERAZAM08/ML-QA.git
cd ML-QA
pip install -r requirements.txt
python app.py