Update app.py

app.py

@@ -1,4 +1,3 @@
-!pip install phonemizer
 
 import gradio as gr
 import numpy as np
@@ -11,14 +10,14 @@ from transformers import VitsModel, VitsTokenizer, FSMTForConditionalGeneration,
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
 
 # Transform audio to en text
-asr_pipe = pipeline("automatic-speech-recognition", model="asapp/sew-d-tiny-100k-ft-ls100h", device=device)
+asr_pipe = pipeline("automatic-speech-recognition", model="asapp/sew-d-tiny-400k-ft-ls100h", device=device)
 
 # Translate en to rus text
 translation_en_to_rus = pipeline("translation", model="Helsinki-NLP/opus-mt-en-ru")
 
 # Create speech from rus text
-model = VitsModel.from_pretrained("Helsinki-NLP/opus-mt-ru-en")
-tokenizer = VitsTokenizer.from_pretrained("Helsinki-NLP/opus-mt-ru-en")
+model = VitsModel.from_pretrained("facebook/mms-tts-rus")
+tokenizer = VitsTokenizer.from_pretrained("facebook/mms-tts-rus")
 
 #model = MarianMTModel.from_pretrained("Helsinki-NLP/opus-mt-ru-en")
 #tokenizer = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-ru-en")