Update handler.py

handler.py

@@ -14,20 +14,21 @@ class EndpointHandler:
         self.tokenizer = AutoTokenizer.from_pretrained(path)
         #self.feature_extractor = AutoFeatureExtractor.from_pretrained(path)
         self.model = ParlerTTSForConditionalGeneration.from_pretrained(path).to(device)  #torch_dtype=torch.float16
+        self.number_normalizer = EnglishNumberNormalizer()  # Initialize number normalizer
 
-        def preprocess_text(self, text):
-            """Implement the same preprocessing as the Gradio app"""
-            text = self.number_normalizer(text).strip()
-            text = text.replace("-", " ")
-            if text[-1] not in punctuation:
-                text = f"{text}."
-            
-            abbreviations_pattern = r'\b[A-Z][A-Z\.]+\b'
-            abbreviations = re.findall(abbreviations_pattern, text)
-            for abv in abbreviations:
-                if abv in text:
-                    text = text.replace(abv, " ".join(abv.replace(".","")))
-            return text
+    def preprocess_text(self, text):
+        """Implement the same preprocessing as the Gradio app"""
+        text = self.number_normalizer(text).strip()
+        text = text.replace("-", " ")
+        if text[-1] not in punctuation:
+            text = f"{text}."
+        
+        abbreviations_pattern = r'\b[A-Z][A-Z\.]+\b'
+        abbreviations = re.findall(abbreviations_pattern, text)
+        for abv in abbreviations:
+            if abv in text:
+                text = text.replace(abv, " ".join(abv.replace(".","")))
+        return text
     
     def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
         """
@@ -62,8 +63,8 @@ class EndpointHandler:
         with torch.autocast(device):
             outputs = self.model.generate(
                 **voice_description, prompt_input_ids=inputs.input_ids, 
-                prompt_attention_mask=inputs.attention_mask, attention_mask=inputs.attention_mask, 
-                **parameters
+                prompt_attention_mask=voice_description.attention_mask, attention_mask=inputs.attention_mask, 
+                **gen_kwargs
             )
 
         # postprocess the prediction