app.py

import gradio as gr
from TTS.api import TTS
import tempfile

# Load viXTTS model
tts = TTS(model_path="ntdgo/ttsvi", progress_bar=False, gpu=True)

def synthesize_tts(audio_file, text):
    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp_output:
        tts.tts_to_file(
            text=text,
            speaker_wav=audio_file,
            language="vi",
            file_path=tmp_output.name
        )
        return tmp_output.name

with gr.Blocks() as demo:
    gr.Markdown("## 🇻🇳 Vietnamese Emotional TTS with Voice Cloning")
    with gr.Row():
        audio_input = gr.Audio(label="🎤 Upload your voice (WAV only)", type="filepath")
        text_input = gr.Textbox(label="📝 Vietnamese Text", value="Xin chào! Tôi rất vui được gặp bạn.")
    output_audio = gr.Audio(label="🗣️ Generated Voice Output")
    run_btn = gr.Button("Generate")
    run_btn.click(fn=synthesize_tts, inputs=[audio_input, text_input], outputs=output_audio)

demo.launch()