from flask import Flask, render_template, request, jsonify, after_this_request from io import BytesIO import base64 import subprocess import os import random import string import re app = Flask(__name__) # Define the folder where files are saved file_folder = '/home/app/' # Models with specific character replacements models_replacements = { "Español México | Kamora Neuronal": { "model_path": "es_ES-kamora-7539-high.onnx", "replacements": [('\n', '. ')] }, "Español México | Claude": { "model_path": "es_MX-claude-14947-epoch-high.onnx", "replacements": [('(', ','), (')', ','), ('?', ','), ('¿', ','), (':', ','), ('\n', ' ')] } } def filter_text(text, replacements): filtered_text = re.sub(r'[^\w\s,.\(\):\u00C0-\u00FF]', '', text) for replacement in replacements: filtered_text = filtered_text.replace(replacement[0], replacement[1]) return filtered_text def convert_text_to_speech(parrafo, model): # Limit text to 500 characters parrafo = parrafo[:100000] model_info = models_replacements.get(model) if model_info: model_path = model_info.get("model_path") replacements = model_info.get("replacements") parrafo_filtrado = filter_text(parrafo, replacements) random_name = ''.join(random.choices(string.ascii_letters + string.digits, k=8)) + '.wav' output_file = os.path.join(file_folder, random_name) app.logger.info("Audio file created at: %s", output_file) piper_exe = os.path.join(file_folder, 'piper') # Adjusted the path for piper if os.path.isfile(piper_exe): comando = f'echo {parrafo_filtrado} | "{piper_exe}" -m {model_path} -f {output_file}' subprocess.run(comando, shell=True) return output_file else: return "The piper.exe file was not found in the correct directory." else: return "Model not found." @app.route('/') def index(): model_options = list(models_replacements.keys()) # Log the contents of the current folder app.logger.info("Contents of current folder: %s", os.listdir(file_folder)) return render_template('index.html', model_options=model_options) @app.route('/convert', methods=['POST']) def convert_text(): text = request.form['text'] model = request.form['model'] output_file = convert_text_to_speech(text, model) @after_this_request def remove_file(response): try: os.remove(output_file) app.logger.info("Audio file deleted: %s", output_file) except Exception as error: app.logger.error("Error deleting file: %s", error) return response with open(output_file, 'rb') as audio_file: audio_content = audio_file.read() audio_base64 = base64.b64encode(audio_content).decode('utf-8') response = jsonify({'audio_base64': audio_base64}) return response if __name__ == '__main__': app.run(host='0.0.0.0', port=7860, debug=False)