app.py

import gradio as gr
from TTS.api import TTS
import torch
import os
from pydub import AudioSegment

# CSS for warnings and styling
css = """
#warning {background-color: #FFCCCB !important}
.feedback label textarea {
    height: auto !important;
    font-size: 22px !important;
    font-weight: 800 !important;
    text-align: center !important;
    color: #801313 !important;
    padding: 0px !important
}
#alert {background-color: #fff !important}
"""

# Use GPU if available
device = "cuda" if torch.cuda.is_available() else "cpu"

# Load models
tts = TTS(model_name="tts_models/multilingual/multi-dataset/your_tts", progress_bar=False, gpu=(device=="cuda"))
tts.to(device)

# Convert mp3 to wav
def convert_mp3_to_wav(mp3_path: str) -> str:
    wav_path = mp3_path.replace(".mp3", ".wav")
    audio = AudioSegment.from_mp3(mp3_path)
    audio.export(wav_path, format="wav")
    return wav_path

# Voice cloning function
def text_to_speech(text: str, speaker_wav: str, speaker_wav_file: str):
    text = text.strip().replace("\n", " ")
    speaker_audio = speaker_wav_file or speaker_wav

    if not text:
        return None, "⚠️ Error: Text input is empty."
    if not speaker_audio or not os.path.exists(speaker_audio):
        return None, "⚠️ Error: No valid speaker audio provided."

    if speaker_audio.endswith(".mp3"):
        try:
            speaker_audio = convert_mp3_to_wav(speaker_audio)
        except Exception as e:
            return None, f"⚠️ Error converting MP3 to WAV: {str(e)}"

    output_path = "output.wav"

    try:
        tts.tts_to_file(text=text, speaker_wav=speaker_audio, language="en", file_path=output_path)
        if os.path.exists(output_path) and os.path.getsize(output_path) > 0:
            return output_path, ""
        else:
            return None, "⚠️ Error: Audio was not generated."
    except Exception as e:
        return None, f"⚠️ Error during synthesis: {str(e)}"

# Toggle mic/file input visibility
def toggle(choice: str):
    return (
        gr.update(visible=(choice == "mic"), value=None),
        gr.update(visible=(choice != "mic"), value=None)
    )

# Change alert style based on input
def change_color(text_input: str):
    return gr.update(elem_id="warning" if len(text_input) == 0 else "alert", autofocus=(len(text_input) == 0))

# Reset fields
def clear_color(text_input: str, radio: str, error_box: str):
    return gr.update(elem_id="alert"), gr.update(value="mic"), gr.update(visible=False)

# Show error or success
def show_error(text: str):
    return (
        gr.update(visible=(text == ""), elem_id="warning", elem_classes="feedback"),
        gr.update(visible=(text != ""))
    )

# Gradio UI
with gr.Blocks(css=css) as demo:
    with gr.Row():
        with gr.Column():
            text_input = gr.Textbox(label="Enter text to clone", value="", max_lines=4, lines=4)
            radio = gr.Radio(["mic", "file"], value="mic", label="Upload speaker audio")
            audio_input_mic = gr.Audio(label="Use Microphone", sources="microphone", type="filepath", visible=True)
            audio_input_file = gr.Audio(label="Upload File (.wav/.mp3)", type="filepath", visible=False)

            with gr.Row():
                with gr.Column():
                    btn_clear = gr.ClearButton([text_input, radio, audio_input_file])
                with gr.Column():
                    btn = gr.Button("Generate Voice", variant="primary")

        with gr.Column():
            audio_output = gr.Audio(label="Generated Voice", visible=True, autoplay=True, show_share_button=False)
            error_box = gr.Textbox(label="Status", value="Input box cannot be blank!!", visible=False, container=True)

    # Event bindings
    btn_clear.add(audio_output)
    btn.click(text_to_speech, inputs=[text_input, audio_input_mic, audio_input_file], outputs=[audio_output, error_box])
    btn.click(show_error, text_input, [error_box, audio_output])
    radio.change(toggle, radio, [audio_input_mic, audio_input_file])
    btn_clear.click(clear_color, [text_input, radio, error_box], [text_input, radio, error_box])
    btn.click(change_color, text_input, text_input)

# Launch the app
demo.launch()