app.py

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

# Configure Google Gemini API
genai.configure(api_key=os.getenv("GEMINI_API_KEY"))

# Play.ht API keys
API_KEY = os.getenv('PLAY_API_KEY')
USER_ID = os.getenv('PLAY_USER_ID')

# Function to upload image to Gemini and get roasted text
def upload_to_gemini(path, mime_type="image/jpeg"):
    file = genai.upload_file(path, mime_type=mime_type)
    return file

def generate_roast(image_path):
    # Upload the image to Gemini and get the text
    uploaded_file = upload_to_gemini(image_path)
    generation_config = {
        "temperature": 1,
        "top_p": 0.95,
        "top_k": 40,
        "max_output_tokens": 8192,
        "response_mime_type": "text/plain",
    }
    model = genai.GenerativeModel(
        model_name="gemini-1.5-flash-002",
        generation_config=generation_config,
        system_instruction="You are a professional satirist and fashion expert. You will be given a profile picture. Your duty is to roast whatever is given to you in the funniest way possible!",
    )
    
    chat_session = model.start_chat(
        history=[{"role": "user", "parts": [uploaded_file]}]
    )
    response = chat_session.send_message("Roast this image!")
    return response.text

# Function to convert text to speech with Play.ht
def text_to_speech(text):
    url = "https://api.play.ht/api/v2/tts/stream"
    payload = {
        "voice": "s3://voice-cloning-zero-shot/d9ff78ba-d016-47f6-b0ef-dd630f59414e/female-cs/manifest.json",
        "output_format": "mp3",
        "text": text,
    }
    headers = {
        "accept": "audio/mpeg",
        "content-type": "application/json",
        "Authorization": API_KEY,
        "X-User-ID": USER_ID
    }
    
    response = requests.post(url, json=payload, headers=headers)
    if response.status_code == 200:
        audio_path = "output_audio.mp3"
        with open(audio_path, "wb") as audio_file:
            audio_file.write(response.content)
        return audio_path
    else:
        return f"Error: {response.status_code} - {response.text}"

# Gradio Interface
with gr.Blocks(theme={"primary_hue": "#b4fd83"}) as demo:
    gr.Markdown("# Image to Text-to-Speech Roasting App")
    gr.Markdown("Upload an image, and the AI will roast it and convert the roast to audio.")
    
    with gr.Row():
        with gr.Column():
            image_input = gr.Image(type="filepath", label="Upload Image")
        with gr.Column():
            output_text = gr.Textbox(label="Roast Text")
            audio_output = gr.Audio(label="Roast Audio")
    
    def process_image(image):
        roast_text = generate_roast(image)
        audio_path = text_to_speech(roast_text)
        return roast_text, audio_path
    
    submit_button = gr.Button("Generate Roast")
    submit_button.click(process_image, inputs=image_input, outputs=[output_text, audio_output])

# Launch the app
demo.launch(debug=True)