Update app.py

app.py

@@ -12,12 +12,14 @@ tts_model = TTS(model_name="tts_models/en/ljspeech/tacotron2-DDC")
 
 # Emotion-specific settings for pitch and speed
 emotion_settings = {
-    "joy": {"pitch": 1.2, "speed": 1.1},
-    "sadness": {"pitch": 0.8, "speed": 0.9},
-    "anger": {"pitch": 1.0, "speed": 1.2},
-    "fear": {"pitch": 0.9, "speed": 1.0},
-    "surprise": {"pitch": 1.3, "speed": 1.2},
     "neutral": {"pitch": 1.0, "speed": 1.0},
+    "joy": {"pitch": 1.3, "speed": 1.2},
+    "sadness": {"pitch": 0.8, "speed": 0.9},
+    "anger": {"pitch": 1.6, "speed": 1.4},
+    "fear": {"pitch": 1.2, "speed": 0.95},
+    "surprise": {"pitch": 1.5, "speed": 1.3},
+    "disgust": {"pitch": 0.9, "speed": 0.95},
+    "shame": {"pitch": 0.8, "speed": 0.85},
 }
 
 
@@ -43,11 +45,10 @@ def emotion_aware_tts_pipeline(input_text=None, file_input=None):
             audio_path = "output.wav"
             tts_model.tts_to_file(text=input_text, file_path=audio_path)
 
-            # Adjust pitch and speed using librosa
-            if pitch != 1.0:
-                adjust_audio_pitch(audio_path, pitch)
-            if speed != 1.0:
-                adjust_audio_speed(audio_path, speed)
+            # Adjust pitch
+            pitch_factor = (pitch - 1.0) * 12  # Convert to semitones for librosa
+            adjust_pitch(audio_path, pitch_factor)
+
 
             return f"Detected Emotion: {emotion} (Confidence: {confidence:.2f})", audio_path
         else:
@@ -56,7 +57,6 @@ def emotion_aware_tts_pipeline(input_text=None, file_input=None):
         return f"Error: {str(e)}", None
 
 
-
 # Define Gradio interface
 interface = gr.Interface(
     fn=emotion_aware_tts_pipeline,