Update app.py

app.py

@@ -12,27 +12,36 @@ VOICES = {
 }
 
 def parse_segments(text):
-    """Parse input text for speaker segments using regex"""
-    pattern = re.compile(r'$$(?P<speaker>[^$$]+)$$(?P<text>.*?)$$\/\1$$', re.DOTALL)
-    return [(match.group('speaker'), match.group('text').strip()) 
-            for match in pattern.finditer(text)]
+    """Parse input text for speaker segments with improved validation"""
+    pattern = re.compile(r'$$(?P<speaker>[^$$]+?)$$(?P<text>.*?)$$/(?P=speaker)$$', re.DOTALL)
+    matches = list(pattern.finditer(text))
+    
+    # Validate speaker names and collect results
+    valid_segments = []
+    for match in matches:
+        speaker = match.group('speaker')
+        if speaker in VOICES:
+            valid_segments.append((speaker, match.group('text').strip()))
+    
+    # Find any invalid segments
+    if len(matches) < len(text.strip()):
+        return valid_segments, f"Warning: Found {len(matches)} valid segments, but text contains untagged content or invalid speaker names"
+    
+    return valid_segments, None
 
 def generate_podcast(input_text):
     """Convert text to podcast with multiple voices"""
     try:
-        segments = parse_segments(input_text)
+        segments, warning = parse_segments(input_text)
         
         if not segments:
-            return (22050, np.zeros(0)), "No valid speaker segments found"
-            
+            return (22050, np.zeros(0)), "No valid speaker segments found. Please use the format: [Speaker Name]text[/Speaker Name]"
+        
         all_audio = []
         current_pipe = None
         current_model = ""
         
         for speaker, text in segments:
-            if speaker not in VOICES:
-                return (22050, np.zeros(0)), f"Invalid speaker: {speaker}"
-                
             model_name = VOICES[speaker]
             
             # Load model only when needed
@@ -48,7 +57,11 @@ def generate_podcast(input_text):
         # Combine all audio segments with short pauses
         final_audio = np.concatenate([np.concatenate((audio, np.zeros(5000))) for audio in all_audio])
         
-        return (output["sampling_rate"], final_audio), "Podcast generated successfully!"
+        status = "Podcast generated successfully!"
+        if warning:
+            status += " " + warning
+            
+        return (output["sampling_rate"], final_audio), status
 
     except Exception as e:
         return (22050, np.zeros(0)), f"Error: {str(e)}"
@@ -64,8 +77,8 @@ demo = gr.Interface(
         label="Input Text with Speaker Tags",
         lines=12,
         placeholder="""Example format:
-[Amy (Female)]Hello and welcome to today's episode![/Amy (Female)]
-[Joe (Male)]Excited to have you here![/Joe (Male)]"""
+[Amy (Female)]Welcome to our podcast![/Amy (Female)]
+[Joe (Male)]Today we're discussing AI innovations.[/Joe (Male)]"""
     ),
     outputs=[
         gr.Audio(label="Generated Podcast", type="numpy"),
@@ -73,7 +86,8 @@ demo = gr.Interface(
     ],
     examples=[
         ["""[Amy (Female)]Welcome to our podcast![/Amy (Female)]
-[Joe (Male)]Today we're discussing AI innovations.[/Joe (Male)]"""]
+[Joe (Male)]Today we're discussing AI innovations.[/Joe (Male)]
+[Clara (Female)]This is Clara speaking![/Clara (Female)]"""]
     ],
     title="🎙️ Multi-Voice Podcast Generator",
     description="Generate podcasts with multiple free AI voices using Microsoft's Piper TTS models. Use [SpeakerName] tags to assign different voices to different text segments.",