Voice Assistant Backend

Overview

This FastAPI backend handles speech-to-text, natural language processing, and text-to-speech conversion using OpenAI's Whisper, GPT models, and Azure's Speech Services.

Setup & Prerequisites

File Structure

.
├── main.py          # Main application file
├── test.py          # Test file generator
├── .env             # Environment variables
├── requirements.txt # Required packages
└── README.md        # This file

Environment Variables

Create a .env file in the root directory with the following:

OPENAI_API_KEY=your_openai_key_here
AZURE_SPEECH_KEY=your_azure_speech_key_here
AZURE_SPEECH_REGION=your_azure_region_here

VoiceAssistant Class

Manages recording, transcription, chat, and playback
Handles conversation state
Coordinates API interactions

FastAPI Integration

Single endpoint: /chat
Supports both streaming and non-streaming responses
Swagger UI available at /docs

Main Functions

record_audio(): Captures microphone input
process_audio(): Converts speech to text using Whisper
get_chat_response(): Generates response using ChatGPT
synthesize_speech(): Converts text to speech
play_audio(): Plays the response

Implementation Flow

User initiates chat through endpoint
System records audio
Audio processed through Whisper
Response generated via ChatGPT
Response converted to speech
Audio played back to user

Testing

Run test file generator: python3 test.py
Navigate to /docs to test the /chat endpoint
Basic error handling tests included
Core functionality tests for each component

Running the Application

Install dependencies:

pip install -r requirements.txt

Set up your .env file with the required API keys
Start the server:

uvicorn main:app --reload

Navigate to http://localhost:8000/docs to test the /chat endpoint
Generate test WAV file:

python3 test.py

License

Silianglei/chatbackend

ErrorLooks like something went wrong!

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Languages

Error
Looks like something went wrong!

Packages