ParrotAI

ParrotAI is a Flask/Python-based application that serves as a proof of concept for generating speech using the coqui/TTS model from HuggingFace. This project is a quick proof of concept designed to explore the use of HuggingFace models for text-to-speech (TTS) synthesis while providing a simple user interface with options for selecting languages, recording voice samples, or submitting .wav files for custom voice generation.

Note: This application is not production-ready and requires many enhancements before it can be considered for production use.

Features

• Text-to-Speech (TTS): Generates speech from text using the HuggingFace coqui/TTS model.
• Language Selection: Users can select from a variety of languages to generate the audio.
• Voice Sample Recording: Users can record their own voice samples in .wav format, which the model will use to generate personalized responses.
• File Upload: Users can upload .wav files to be used for generating audio responses.
• Default Audio: If no custom voice samples are provided, the application will use a standard female.wav file to generate audio.
• Responsive Frontend: The frontend is styled using Tailwind CSS (from the CDN) and provides a user-friendly interface for selecting languages, recording voice samples, and submitting files.

Prerequisites

• Python 3.10.12: This application requires Python version 3.10.12 to run properly.
• HuggingFace Account: The application connects to HuggingFace to use the coqui/TTS model, so you'll need access to HuggingFace models.

Installation

1. Clone the Repository:

   git clone https://github.com/dantasl/parrot-ai.git
   cd ParrotAI

2. Set up a Virtual Environment:

   python3.10 -m venv venv
   source venv/bin/activate

3. Install Required Dependencies:

   pip install -r requirements.txt

Usage

Start the Flask Application:

flask run

Access the Application:

Once the Flask server is running, you can access the application in your browser at:

http://127.0.0.1:5000/

Language Selection:

• Select the language from the dropdown menu (e.g., English, French, Spanish, etc.).
• Type your message in the textarea and submit it.

Recording a Voice Sample:

• Click on the "Record" button to start recording your voice.
• After recording, click "Stop" to save the sample, which will be used to generate audio for subsequent requests.

Upload a .wav File:

• Use the file input to upload a .wav file.
• The application will use the latest uploaded file to generate the audio response.

Audio Generation:

• The application will generate a .wav audio file based on the latest voice sample submitted.
• If no voice sample is available, a default female.wav file will be used.

Directory Structure:

ParrotAI/
│
├── audio_outputs/            # Directory where generated .wav files are stored
├── audio_inputs/             # Directory where the uploaded .wav files are stored
├── templates/
│   └── index.html            # Frontend HTML template with Tailwind CSS
├── app.py                    # Main Flask application
├── requirements.txt          # Python dependencies
└── README.md                 # Project README file

Enhancements Required for Production

While ParrotAI is a functional proof of concept, several improvements would be needed to bring it to production:

• Error Handling: More robust error handling and logging for both the frontend and backend.
• Database Integration: Store user interactions, uploaded files, and language preferences in a database.
• Authentication: Implement user authentication and authorization.
• Security Enhancements: Secure file uploads, CSRF protection, and other Flask security best practices.
• Model Optimization: Optimize the TTS model for performance and scale it for higher loads.
• Testing: Add unit and integration tests to ensure application stability.

Technology Stack

• Backend: Flask (Python)
• Frontend: HTML, Tailwind CSS (via CDN)
• Text-to-Speech: HuggingFace coqui/TTS model
• Audio Handling: .wav file recording and uploading