app.py

import os
import re
import unicodedata

import gradio as gr
from cached_path import cached_path
from deep_phonemizer.phonemizer import Phonemizer
from g2p import make_g2p
from huggingface_hub import login

login(token=os.getenv("login"))
normalizer = make_g2p("str", "str-equiv")
checkpoint_path = cached_path("hf://lobot-penac/g2p/model_step_420k.pt")
phonemizer = Phonemizer.from_checkpoint(checkpoint_path)

# Your custom G2P function
def g2p_function(text):
    text = unicodedata.normalize("NFC", normalizer(text).output_string)
    texts = text.split()
    result = phonemizer.phonemise_list(texts, lang="str")
    return " ".join(result.phonemes)


# Gradio interface
with gr.Blocks(theme=gr.themes.Default(font=[gr.themes.GoogleFont("Tinos"), "Arial", "sans-serif"])) as demo:
    gr.Markdown("""
        <style>
        body, textarea, input {
            font-size: 24px !important;
        }
        </style>
    """)
    gr.Markdown("# G2P Converter\nEnter SENĆOŦEN text and see its phonemic transcription.")
    input_text = gr.Textbox(lines=2, label="SENĆOŦEN Orthography", placeholder="Enter text here...")
    output_text = gr.Textbox(label="Phonemes")
    input_text.change(fn=g2p_function, inputs=input_text, outputs=output_text)

# Run the app
if __name__ == "__main__":
    demo.launch(share=True)