Update app.py

app.py

@@ -4,13 +4,16 @@ from scipy.io.wavfile import write
 import gradio as gr
 from transformers import VitsTokenizer, VitsModel, set_seed, pipeline
 from numToLez import numToLez
+import spaces
 
 # Load your fine-tuned model
 model_name = "leks-forever/vits_lez_tts"  # Replace with your Hugging Face model name
 tokenizer = VitsTokenizer.from_pretrained(model_name)
 model = VitsModel.from_pretrained(model_name)
+model.to("cuda")
 
 tts_pipeline = pipeline("text-to-speech", model=model_name)
+tts_pipeline.to("cuda")
 
 new_sentence = '!.?'
 in_sentence = ',-.:;'
@@ -22,7 +25,7 @@ def canonize_lez(text):
             text = text.replace(abruptive_letter+abruptive_symbol, abruptive_letter+'Ӏ')
     return text
 
-
+@spaces.GPU()
 def tts_function(input_text, speaking_rate, noise_scale, add_pauses):
     fixed_text = canonize_lez(input_text)
     if add_pauses: