app.py

import torch
import re
import gradio as gr
import soundfile as sf
import numpy as np
from transformers import SpeechT5HifiGan
from IPython.display import Audio
from transformers import SpeechT5ForTextToSpeech
from transformers import SpeechT5Processor

# helper function
number_words = {
    0: "zero", 1: "one", 2: "two", 3: "three", 4: "four", 5: "five", 6: "six", 7: "seven", 8: "eight", 9: "nine",
    10: "ten", 11: "eleven", 12: "twelve", 13: "thirteen", 14: "fourteen", 15: "fifteen", 16: "sixteen", 17: "seventeen",
    18: "eighteen", 19: "nineteen", 20: "twenty", 30: "thirty", 40: "forty", 50: "fifty", 60: "sixty", 70: "seventy",
    80: "eighty", 90: "ninety", 100: "hundred", 1000: "thousand"
}
replacements = [
    ("“", '"'),
    ("”", '"'),
    ("’", ","),
    ("_", " "),
    ("\xa0", " "),
    ("\n", " "),
    ("$","dollar"),
    ("%","percent"),
    ("&","and"),
    ("*","star"),
    ("+","plus"),
    ("—","-")

]
def number_to_words(number):
    if number < 20:
        return number_words[number]
    elif number < 100:
        tens, unit = divmod(number, 10)
        return number_words[tens * 10] + (" " + number_words[unit] if unit else "")
    elif number < 1000:
        hundreds, remainder = divmod(number, 100)
        return (number_words[hundreds] + " hundred" if hundreds > 1 else "hundred") + (" " + number_to_words(remainder) if remainder else "")
    elif number < 1000000:
        thousands, remainder = divmod(number, 1000)
        return (number_to_words(thousands) + " thousand" if thousands > 1 else "thousand") + (" " + number_to_words(remainder) if remainder else "")
    elif number < 1000000000:
        millions, remainder = divmod(number, 1000000)
        return number_to_words(millions) + " million" + (" " + number_to_words(remainder) if remainder else "")
    elif number < 1000000000000:
        billions, remainder = divmod(number, 1000000000)
        return number_to_words(billions) + " billion" + (" " + number_to_words(remainder) if remainder else "")
    else:
        return str(number)

def replace_numbers_with_words(text):

    def replace(match):
        number = int(match.group())
        return number_to_words(number)

    # Find the numbers and change with words.
    result = re.sub(r'\b\d+\b', replace, text)

    return result

def cleanup_text(text):
    for src, dst in replacements:
        text = text.replace(src, dst)
    return text

def normalize_text(text):
    # Convert to lowercase
    text = text.lower()

    # Remove punctuation (except apostrophes)
    text = re.sub(r'[^\w\s\']', '', text)

    # Remove extra whitespace
    text = ' '.join(text.split())

    return text

model = SpeechT5ForTextToSpeech.from_pretrained(
    "Yassmen/speecht5_finetuned_english_tehnical"
)

checkpoint = "microsoft/speecht5_tts"
processor = SpeechT5Processor.from_pretrained(checkpoint)



def generate_wav_file(text):
    try:  
      converted_text = replace_numbers_with_words(text)
      cleaned_text = cleanup_text(converted_text)
      final_text = normalize_text(cleaned_text)
      inputs = processor(text=final_text, return_tensors="pt")
      speaker_embeddings = torch.tensor(np.load('speaker_embedding.npy'))
      vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan")
      speech = model.generate_speech(inputs["input_ids"], speaker_embeddings, vocoder=vocoder)
    # Convert the speech to a WAV file
      output_file = "output.wav"
      sf.write(output_file, speech.detach().cpu().numpy(), 16000)
      return output_file  # Return the file path for download
    except Exception as e:
        print(f"Error: {e}")
        return None

iface = gr.Interface(
    fn=generate_wav_file,
    inputs=gr.Textbox(lines=3, label="Enter text to convert to speech"),
    outputs= gr.Audio(type="filepath", label="Generated Audio"),
    title="Text-to-Speech Technical EN"
)
if __name__ == "__main__":
  iface.launch()