import streamlit as st
from transformers import pipeline
import soundfile as sf
import tempfile

# Streamlit page setup
st.set_page_config(page_title="Hugging Face Text-to-Speech", page_icon="🎤", layout="centered")
st.title("🎤 Text to Speech Converter (Hugging Face)")

# Text input from user
text = st.text_area("Enter the text you want to convert to speech:", height=150)

# Choose model
model_name = st.selectbox(
    "Select a Hugging Face TTS model:",
    ["suno/bark-small", "facebook/speecht5_tts", "espnet/kan-bayashi_ljspeech_vits"],
    index=0
)

if st.button("Convert to Speech"):
    if text.strip() == "":
        st.warning("Please enter some text.")
    else:
        st.info("Generating speech... please wait ⏳")
        tts = pipeline("text-to-speech", model=model_name)
        output = tts(text)

        # Get audio and sampling rate
        audio = output["audio"]
        sr = output["sampling_rate"]

        # Save audio to a temporary file
        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmpfile:
            sf.write(tmpfile.name, audio, sr)
            audio_path = tmpfile.name

        st.success("✅ Conversion complete!")
        st.audio(audio_path, format="audio/wav")
        st.download_button(
            label="📥 Download Audio",
            data=open(audio_path, "rb").read(),
            file_name="tts_output.wav",
            mime="audio/wav"
        )