app.py

import streamlit as st
import gradio as gr
from speechbrain.pretrained import Tacotron2, HIFIGAN
from scipy.io.wavfile import write

# Load TTS and vocoder models
tacotron2 = Tacotron2.from_hparams(source="speechbrain/tts-tacotron2-ljspeech", savedir="tmpdir_tts")
hifi_gan = HIFIGAN.from_hparams(source="speechbrain/tts-hifigan-ljspeech", savedir="tmpdir_vocoder")

# Function to generate speech from text
def text_to_speech(text):
    mel_output, mel_length, alignment = tacotron2.encode_text(text)
    waveforms, _, _ = hifi_gan.decode_batch(mel_output)
    audio_path = "tts_output.wav"
    write(audio_path, 22050, waveforms.squeeze(1).cpu().numpy())
    return audio_path

# Gradio interface for TTS
def chatbot_response(text):
    audio_path = text_to_speech(text)
    return audio_path

gr_interface = gr.Interface(fn=chatbot_response, inputs="text", outputs="audio")

# Streamlit app setup
st.title("Text-to-Speech Chatbot")
st.write("This application converts text into speech using SpeechBrain.")

# Embed Gradio interface in Streamlit
with st.spinner("Loading Gradio interface..."):
    gr_interface.launch(share=False, inbrowser=True)