add

app.py

@@ -1,3 +1,21 @@
 import gradio as gr
+import scipy
+from transformers import VitsModel, AutoTokenizer
+import torch
 
-gr.Interface.load("models/facebook/mms-tts-crh").launch()
+model = VitsModel.from_pretrained("facebook/mms-tts-crh")
+tokenizer = AutoTokenizer.from_pretrained("facebook/mms-tts-crh")
+
+def greet(name):
+    global model, tokenizer
+    text = name
+    
+    inputs = tokenizer(text, return_tensors="pt")
+
+    with torch.no_grad():
+        output = model(**inputs).waveform
+
+    return output
+
+iface = gr.Interface(fn=greet, inputs="text", outputs="audio")
+iface.launch()