pirate-bot

Voice to Voice chatbot using Whisper + Open AI + 11labs

Voice-to-Voice Chatbot

A voice-to-voice chatbot powered by OpenAI GPT-3.5 Turbo and ElevenLabs. The chatbot transcribes the audio input using OpenAI's Whisper ASR system, generates a response with GPT-3.5 Turbo, and then converts the response into spoken audio using the ElevenLabs Text-to-Speech API.

Getting Started

1. Clone this repository.
2. Install the required packages by running pip install -r requirements.txt (you should have Python 3.6 or later installed).
3. Create a .env file in the root directory of this project and set your OpenAI API key and ElevenLabs API key.

Example .env file:

Running the Server

You can start the Flask server by running the command: python app.py. This will start the server on http://127.0.0.1:5000/.

Endpoints

• GET / - Render the index page.
• POST /transcribe - Transcribe the given audio to text using Whisper. Requires a file parameter with the file data.
• POST /ask - Generate a ChatGPT response from the given conversation, then convert it to audio using ElevenLabs. Requires a conversation parameter with the conversation data.
• GET /listen/<filename> - Return the audio file located at the given filename.

License

This project is licensed under the terms of the MIT license.