diff --git a/rotary_phone_web.py b/rotary_phone_web.py
index 1eb4377..416c058 100644
--- a/rotary_phone_web.py
+++ b/rotary_phone_web.py
@@ -228,8 +228,9 @@ class RotaryPhone:
             # Play the sound with volume control
             data = wf.readframes(CHUNK)
             while data:
-                # For greeting sounds, stop if handset is hung up
-                if check_hook_status and self.phone_status != "off_hook":
+                # For greeting sounds, stop immediately if handset is hung up
+                if check_hook_status and GPIO.input(HOOK_PIN) != HOOK_PRESSED:
+                    print("Handset hung up, stopping playback immediately")
                     break
 
                 # Apply volume by converting to numpy array and scaling
@@ -256,9 +257,11 @@ class RotaryPhone:
         timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
         filename = os.path.join(OUTPUT_DIR, f"recording_{timestamp}.wav")
         self.current_recording = filename
-        
+
         print(f"Recording to {filename}")
-        
+
+        MIN_RECORDING_DURATION = 1.0  # Minimum 1 second to save recording
+
         try:
             # Use configured audio device
             stream = self.audio.open(
@@ -269,44 +272,56 @@ class RotaryPhone:
                 input_device_index=AUDIO_DEVICE_INDEX,
                 frames_per_buffer=CHUNK
             )
-            
+
             frames = []
             self.recording = True
-            
+
             # Record until handset is hung up or max time reached
             start_time = time.time()
             while self.recording and (time.time() - start_time) < RECORD_SECONDS:
-                # Check if handset is still off hook
+                # Check if handset is still off hook - immediate detection
                 if GPIO.input(HOOK_PIN) != HOOK_PRESSED:
-                    print("Handset hung up, stopping recording")
+                    print("Handset hung up, stopping recording immediately")
                     break
-                
+
                 try:
                     data = stream.read(CHUNK, exception_on_overflow=False)
                     frames.append(data)
                 except Exception as e:
                     print(f"Error reading audio: {e}")
                     break
-            
+
             stream.stop_stream()
             stream.close()
-            
-            # Save the recording
-            if frames:
+
+            # Calculate recording duration
+            duration = len(frames) * CHUNK / RATE if frames else 0
+
+            # Only save if recording is long enough
+            if frames and duration >= MIN_RECORDING_DURATION:
                 wf = wave.open(filename, 'wb')
                 wf.setnchannels(CHANNELS)
                 wf.setsampwidth(self.audio.get_sample_size(FORMAT))
                 wf.setframerate(RATE)
                 wf.writeframes(b''.join(frames))
                 wf.close()
-                duration = len(frames) * CHUNK / RATE
                 print(f"Recording saved: {filename} ({duration:.1f}s)")
             else:
-                print("No audio recorded")
-                
+                # Delete aborted/too short recording
+                if os.path.exists(filename):
+                    os.remove(filename)
+                print(f"Recording aborted or too short ({duration:.1f}s), not saved")
+
         except Exception as e:
             print(f"Recording error: {e}")
-        
+            # Clean up failed recording file
+            if os.path.exists(filename):
+                try:
+                    os.remove(filename)
+                    print(f"Cleaned up failed recording: {filename}")
+                except:
+                    pass
+
         self.recording = False
         self.current_recording = None