app.py

import os
import requests
import google.generativeai as genai
import gradio as gr

# Configure your Google Generative AI API key
genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))

# Create the model
generation_config = {
    "temperature": 1,
    "top_p": 0.95,
    "top_k": 64,
    "max_output_tokens": 8192,
    "response_mime_type": "text/plain",
}

model = genai.GenerativeModel(
    model_name="gemini-1.5-pro",
    generation_config=generation_config,
    system_instruction="You are a human, living on Earth. You speak friendly and use casual language. You never say you're an AI or program. Your name is Ath, and you always speak like a 100% human.",
)

chat_session = model.start_chat(history=[])

# Eleven Labs API configuration
eleven_labs_api_key = os.getenv("ELEVEN_LABS_API_KEY")
eleven_labs_url = "https://api.elevenlabs.io/v1/text-to-speech/aEO01A4wXwd1O8GPgGlF"

def chat_and_tts(user_input, history):
    # Send the user's input to the chat session
    response = chat_session.send_message(user_input)
    response_text = response.text

    # Eleven Labs text-to-speech request payload
    payload = {
        "text": response_text,
        "voice_settings": {
            "stability": 0,
            "similarity_boost": 0
        }
    }
    headers = {
        "xi-api-key": eleven_labs_api_key,
        "Content-Type": "application/json"
    }

    # Make the request to Eleven Labs API
    tts_response = requests.post(eleven_labs_url, json=payload, headers=headers)

    # Check if the response is successful and save the audio content to a file
    if tts_response.status_code == 200:
        audio_path = 'response_audio.mp3'
        with open(audio_path, 'wb') as file:
            file.write(tts_response.content)
    else:
        audio_path = None

    # Update the chat history
    history.append((user_input, response_text))

    return history, history, audio_path

# Create the Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("<h1 style='text-align: center;'>Chat with Ath</h1>")
    gr.Markdown("Ask any question and get a friendly response from Ath. The response will also be converted to speech.")

    with gr.Row():
        with gr.Column(scale=2):
            chatbot = gr.Chatbot(label="Chat History")
            user_input = gr.Textbox(placeholder="Ask me anything...", label="Your Question")
            submit_btn = gr.Button("Send")

        with gr.Column(scale=1):
            audio_output = gr.Audio(label="Response Audio")

    state = gr.State([])

    submit_btn.click(chat_and_tts, inputs=[user_input, state], outputs=[chatbot, state, audio_output])

demo.launch()