Update app.py

app.py

@@ -142,17 +142,30 @@ class VoiceSynthesizer:
         if self.reference_voice is not None:
             # Use saved reference voice file
             history_prompt = self.reference_voice
-        elif voice_preset:
-            # Use predefined voice preset
-            history_prompt = voice_presets[0] if "v2/en_speaker" not in voice_preset else voice_preset
+        
+        # If no reference voice, use preset
+        if history_prompt is None and voice_preset:
+            # Extract the actual preset value
+            if isinstance(voice_preset, str):
+                # Remove any additional text in parentheses
+                preset_value = voice_preset.split(' ')[0]
+                history_prompt = preset_value if preset_value in voice_presets else voice_presets[0]
+            else:
+                history_prompt = voice_presets[0]
         
         # Generate audio with or without history prompt
         try:
+            # Attempt generation with different approaches
             if history_prompt:
-                audio_array = generate_audio(
-                    text, 
-                    history_prompt=history_prompt
-                )
+                try:
+                    audio_array = generate_audio(
+                        text, 
+                        history_prompt=history_prompt
+                    )
+                except Exception as preset_error:
+                    print(f"Error with specific history prompt: {preset_error}")
+                    # Fallback to default generation
+                    audio_array = generate_audio(text)
             else:
                 # Fallback to default generation
                 audio_array = generate_audio(text)
@@ -254,11 +267,12 @@ def create_interface():
                 with gr.Row():
                     bark_preset = gr.Dropdown(
                         choices=[
-                            "v2/en_speaker_6 (Female)",
-                            "v2/en_speaker_3 (Male)", 
-                            "v2/en_speaker_9 (Neutral)"
+                            "v2/en_speaker_6 (Female Voice)",
+                            "v2/en_speaker_3 (Male Voice)", 
+                            "v2/en_speaker_9 (Neutral Voice)"
                         ],
                         label="Bark Voice Preset",
+                        value="v2/en_speaker_6 (Female Voice)",
                         visible=True
                     )
                     
@@ -311,6 +325,10 @@ def create_interface():
             # Select appropriate preset
             preset = bark_preset if "bark" in model else speecht5_preset
             
+            # Extract preset value if it's a string with additional info
+            if isinstance(preset, str):
+                preset = preset.split(' ')[0]
+            
             return synthesizer.generate_speech(
                 text, 
                 model_name=model_map[model],