app.py

import streamlit as st
import json
from openai import OpenAI
from pathlib import Path
import tempfile
import os

# Initialize OpenAI client with hardcoded API key
# WARNING: This is not secure and should only be used for testing
client = os.getenv("openai")

# System prompt for GPT-4
SYSTEM_PROMPT = """
You are an AI assistant that converts user commands into structured JSON data. Your task is to interpret the user's intent and create a JSON object with relevant fields. Always include a 'response' field in the JSON with a friendly, conversational reply to the user's command.

For example:
User: "Play Taylor Swift on Spotify"
JSON: {
    "action": "play music",
    "application": "Spotify",
    "artist": "Taylor Swift",
    "response": "Sure, I'm playing Taylor Swift on Spotify for you now. Enjoy the music!"
}

User: "What's the weather like in New York?"
JSON: {
    "action": "check weather",
    "location": "New York",
    "response": "I'm checking the weather in New York for you. One moment please."
}

Adapt the JSON structure to fit the context of the command, but always include the 'response' field.
"""

def process_command(command):
    text=command
    response = client.chat.completions.create(
        model="gpt-4o",
        messages=[
            {"role": "system", "content": SYSTEM_PROMPT},
            {"role": "user", "content": text}
        ]
    )
    return json.loads(response.choices[0].message.content)

def text_to_speech(text):
    speech_file_path = Path(tempfile.gettempdir()) / "speech.mp3"
    response = client.audio.speech.create(
        model="tts-1",
        voice="nova",
        input=text
    )
    response.stream_to_file(speech_file_path)
    return speech_file_path

st.title("Text Command Assistant")

# Text input for user command
command = st.text_input("Enter your command:")

if st.button("Process Command"):
    if command:
        # Process the text command
        with st.spinner("Processing command..."):
            result = process_command(command)
        
        # Display JSON result
        st.json(result)

        # Extract response and convert to speech
        response_text = result["response"]
        with st.spinner("Generating speech..."):
            speech_file = text_to_speech(response_text)

        # Play the generated speech
        st.audio(str(speech_file))
    else:
        st.warning("Please enter a command.")

st.markdown("---")
st.write("Created by Shrestha Singh")