Update app.py

app.py

@@ -11,6 +11,7 @@ from datasets import load_dataset
 from huggingface_hub import hf_hub_download
 import uuid
 import wave
+import io
 
 
 from piper import PiperVoice
@@ -138,6 +139,8 @@ def text_to_speech(text, audio_file=None):
     # Split text while preserving "..." (ellipsis)
     segments = re.split(r'(\.\.\.|[\n.])', normalized_text)
 
+    segments = [x.strip() for x in segments]
+
     print("segments: ", segments)
     
     # Merge back the ellipsis with previous segment
@@ -151,6 +154,8 @@ def text_to_speech(text, audio_file=None):
             if temp_segment:
                 combined_segments.append(temp_segment.strip())
             temp_segment = ""
+        elif segment.strip() == "":
+            temp_segment = ""
         else:
             if temp_segment:
                 combined_segments.append(temp_segment.strip())
@@ -167,6 +172,9 @@ def text_to_speech(text, audio_file=None):
     combined_audio = np.array([], dtype=np.int16)
 
     for segment in combined_segments:
+        if segment.strip() == "":
+            continue
+            
         with io.BytesIO() as buffer:
             voice.synthesize(segment, buffer, **synthesize_args)
             buffer.seek(0)