Update app.py

app.py

@@ -59,21 +59,15 @@ def compute_snr(original, enhanced):
     return snr
 
 def vad_plot(y, sr, title):
-    # webrtcvad requires 16-bit mono PCM, sample rate 16000, 10/20/30 ms chunks
-    import webrtcvad
-    import numpy as np
-
     vad = webrtcvad.Vad(2)
     
     if sr != 16000:
-        import librosa
         y = librosa.resample(y, orig_sr=sr, target_sr=16000)
         sr = 16000
 
-    frame_duration_ms = 30  # Can be 10, 20, or 30
-    frame_size = int(sr * frame_duration_ms / 1000)  # samples per frame
+    frame_duration_ms = 30
+    frame_size = int(sr * frame_duration_ms / 1000)
 
-    # Pad signal to be multiple of frame_size
     if len(y) % frame_size != 0:
         pad_len = frame_size - (len(y) % frame_size)
         y = np.pad(y, (0, pad_len))
@@ -85,11 +79,21 @@ def vad_plot(y, sr, title):
         pcm = (frame * 32767).astype(np.int16).tobytes()
         try:
             voiced.append(vad.is_speech(pcm, sr))
-        except Exception as e:
-            print("VAD error:", e)
+        except Exception:
             voiced.append(False)
 
-    return voiced
+    # Plot the VAD decisions
+    plt.figure(figsize=(10, 1.5))
+    plt.plot(voiced, drawstyle='steps-mid')
+    plt.title(title)
+    plt.xlabel("Frame Index")
+    plt.ylabel("Speech")
+    plt.tight_layout()
+    buf = io.BytesIO()
+    plt.savefig(buf, format='png')
+    plt.close()
+    buf.seek(0)
+    return buf
 
 def compute_pesq_mfcc_stoi(original_path, enhanced_path):
     sr = 16000