Update app.py

app.py

@@ -5,6 +5,9 @@ import nltk
 from nltk.corpus import cmudict
 from scipy.io.wavfile import write
 
+# --- FIX 1: Define sample_rate as a global constant ---
+SAMPLE_RATE = 22050
+
 # Download required NLTK data
 nltk.download('averaged_perceptron_tagger')
 nltk.download('cmudict')
@@ -37,18 +40,20 @@ def preprocess_text(text):
     return input_data
 
 # Convert model output to an audio file
-def convert_to_audio(model_output, filename="output.wav", sample_rate=22050):
+def convert_to_audio(model_output, filename="output.wav"):
+    # Now uses the global SAMPLE_RATE constant
     normalized_output = np.interp(model_output, (model_output.min(), model_output.max()), (-1, 1))
-    write(filename, sample_rate, normalized_output.astype(np.float32))
+    write(filename, SAMPLE_RATE, normalized_output.astype(np.float32))
     return filename
 
 # Define function to generate sound effect
-def generate_sfx(text, duration=30):
+def generate_sfx(text, duration): # duration no longer needs a default here
     input_data = preprocess_text(text)
     prediction = model.predict(input_data)
 
     # Generate longer output by repeating or padding
-    audio_data = np.tile(prediction.flatten(), (duration * sample_rate // len(prediction.flatten()) + 1))[:duration * sample_rate]
+    # This line now works because SAMPLE_RATE is defined globally
+    audio_data = np.tile(prediction.flatten(), (duration * SAMPLE_RATE // len(prediction.flatten()) + 1))[:duration * SAMPLE_RATE]
     
     audio_file = convert_to_audio(audio_data, filename="output.wav")
     
@@ -59,7 +64,8 @@ interface = gr.Interface(
     fn=generate_sfx,
     inputs=[
         gr.Textbox(label="Enter a Word", placeholder="Write a Word To Convert it into SFX Sound"),
-        gr.Slider(minimum=2, maximum=20, label="Duration (seconds)", value=30)
+        # --- FIX 2: Corrected the default slider value to be within the min/max range ---
+        gr.Slider(minimum=2, maximum=20, value=5, label="Duration (seconds)")
     ],
     outputs=gr.Audio(label="Generated SFX", type="filepath"),
     title="SFX Generator from Text",
@@ -69,4 +75,4 @@ interface = gr.Interface(
 # Run the interface
 if __name__ == "__main__":
     tf.config.set_visible_devices([], 'GPU')  # Disable GPU
-    interface.launch()
+    interface.launch()