Simple Voice Assistant App

A FastAPI-based voice assistant that processes audio input, transcribes it using OpenAI's Whisper, generates responses using GPT-4, and converts the response back to speech using OpenAI's TTS.

Features

• Audio file upload and processing
• Speech-to-text transcription using OpenAI Whisper
• AI-powered responses using GPT-4
• Text-to-speech conversion using OpenAI TTS
• CORS-enabled for frontend integration

Prerequisites

• Python 3.7 or higher
• OpenAI API key
• pip (Python package installer)

Installation

1. Clone the repository

   git clone <repository-url>
   cd simple-voice-assistant-app

2. Create a virtual environment (recommended)

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

   pip install -r requirements.txt

4. Set up environment variables Or export it directly in your terminal:

   export OPEN_AI_KEY=your_openai_api_key_here

Running the Application

1. Start the FastAPI server

   uvicorn main:app --reload --host 0.0.0.0 --port 8000

2. The API will be available at

   • Local: http://localhost:8000
   • API Documentation: http://localhost:8000/docs (Swagger UI)
   • Alternative docs: http://localhost:8000/redoc

API Endpoints

POST /api/process-audio

Processes an audio file and returns a speech response.

Request:

• Method: POST
• Content-Type: multipart/form-data
• Body: audio file (WAV format recommended)

Response:

• Content-Type: audio/mp3
• Body: MP3 audio file containing the AI's speech response

Example using curl:

curl -X POST "http://localhost:8000/api/process-audio" \
     -H "accept: audio/mp3" \
     -H "Content-Type: multipart/form-data" \
     -F "audio=@your_audio_file.wav"

Usage Example

1. Record an audio file (WAV format) with your question or request
2. Send a POST request to /api/process-audio with the audio file
3. Receive an MP3 file with the AI's spoken response

Frontend Integration

The API is configured with CORS to allow requests from http://localhost:3000. To integrate with a frontend:

1. Set up your frontend to run on port 3000
2. Send audio files to http://localhost:8000/api/process-audio
3. Handle the returned MP3 audio response

Configuration

• CORS Origins: Currently set to http://localhost:3000. Modify line 14 in main.py to add additional origins.
• TTS Voice: Currently using "alloy" voice. Available options: alloy, echo, fable, onyx, nova, shimmer
• AI Model: Using GPT-4 for responses. You can modify line 33 in main.py to use different models.

Troubleshooting

1. OpenAI API Key Issues

   • Ensure your API key is correctly set in the environment variable
   • Verify you have sufficient credits in your OpenAI account

2. Audio Format Issues

   • The app expects audio files in WAV format
   • Ensure your audio file is not corrupted

3. CORS Issues

   • If you're running the frontend on a different port, update the CORS configuration in main.py

Dependencies

• FastAPI: Web framework for building APIs
• OpenAI: Python client for OpenAI API
• Uvicorn: ASGI server for running FastAPI (included in requirements)

License

This project is open source and available under the MIT License.