app.py

import streamlit as st
from io import BytesIO
from urllib.request import urlopen
import librosa
from transformers import Qwen2AudioForConditionalGeneration, AutoProcessor
import pyttsx3  # For text-to-speech

# Load Qwen2Audio model and processor
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-Audio-7B-Instruct")
model = Qwen2AudioForConditionalGeneration.from_pretrained("Qwen/Qwen2-Audio-7B-Instruct", device_map="auto")
tts_engine = pyttsx3.init()

# Streamlit app UI
st.title("Text-to-Audio App")
st.text("This app generates audio from text input using Hugging Face models.")

# User input
text_input = st.text_area("Enter some text for the model:")
if st.button("Generate Audio"):
    conversation = [{"role": "user", "content": text_input}]

    # Preprocess conversation
    text = processor.apply_chat_template(conversation, add_generation_prompt=True, tokenize=False)
    inputs = processor(text=text, return_tensors="pt", padding=True)
    inputs.input_ids = inputs.input_ids.to("cuda")

    # Generate response
    generate_ids = model.generate(**inputs, max_length=256)
    generate_ids = generate_ids[:, inputs.input_ids.size(1):]

    # Decode response
    response = processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
    st.text(f"Model Response: {response}")

    # Convert response to speech
    tts_engine.say(response)
    tts_engine.runAndWait()
    st.success("Audio generated and played!")