diff --git a/src/dsp/pipeline.py b/src/dsp/pipeline.py
index ce025de..1fab85c 100644
--- a/src/dsp/pipeline.py
+++ b/src/dsp/pipeline.py
@@ -352,14 +352,23 @@ class AudioPipeline:
         # load_preset() holds it only for the atomic swap (not during model I/O).
         self._lock = threading.Lock()
 
-        # ── DC-blocking filter state ───────────────────────────────────────
-        # Single-pole high-pass at ~15Hz to remove DC offset and subsonic
-        # noise that can amplify into audible hum through NAM models.
-        # Filter: y[n] = alpha * (y[n-1] + x[n] - x[n-1])
-        # where alpha = 1 / (1 + 2*pi*fc/sr)
-        self._dc_prev_x: float = 0.0
-        self._dc_prev_y: float = 0.0
-        self._dc_alpha: float = 1.0 / (1.0 + 2.0 * np.pi * 15.0 / 48000.0)
+        # ── Mains hum notch filter state (60Hz notch) ───────────────────────
+        # Biquad notch filter to remove 60Hz mains hum without affecting
+        # guitar frequencies (lowest open string is 82Hz E2).
+        # Direct Form I state: two input delays (x1, x2) and two output delays (y1, y2)
+        self._notch_x1: float = 0.0
+        self._notch_x2: float = 0.0
+        self._notch_y1: float = 0.0
+        self._notch_y2: float = 0.0
+        self._notch_b0: float = 1.0
+        self._notch_b1: float = 0.0
+        self._notch_b2: float = 0.0
+        self._notch_a1: float = 0.0
+        self._notch_a2: float = 0.0
+        # Compute initial coefficients for 60Hz notch with Q=5
+        _b0, _b1, _b2, _a1, _a2 = _compute_notch_coeffs(60.0, 5.0, 48000.0)
+        self._notch_b0, self._notch_b1, self._notch_b2 = _b0, _b1, _b2
+        self._notch_a1, self._notch_a2 = _a1, _a2
 
         logger.info("Audio pipeline initialized (block=%d, sr=%d)",
                      self._block_size, self._sample_rate)
@@ -655,20 +664,25 @@ class AudioPipeline:
         )
 
         buf = audio_in.copy()
-        # ── DC-blocking high-pass filter ────────────────────────────
-        # Removes DC offset and subsonic noise (~15Hz corner) that can
-        # amplify into audible hum through high-gain NAM models.
-        # First-order HPF: y[n] = alpha * (y[n-1] + x[n] - x[n-1])
-        alpha = self._dc_alpha
-        px = self._dc_prev_x
-        py = self._dc_prev_y
+        # ── 60Hz mains hum notch filter ─────────────────────────────
+        # Biquad notch at 60Hz (Q=5) to remove mains hum without
+        # affecting guitar frequencies (lowest open string is 82Hz E2).
+        b0, b1, b2 = self._notch_b0, self._notch_b1, self._notch_b2
+        a1, a2 = self._notch_a1, self._notch_a2
+        x1, x2 = self._notch_x1, self._notch_x2
+        y1, y2 = self._notch_y1, self._notch_y2
         for i in range(len(buf)):
-            y = alpha * (py + buf[i] - px)
-            px = buf[i]
-            py = y
+            x = buf[i]
+            y = b0*x + b1*x1 + b2*x2 - a1*y1 - a2*y2
+            x2 = x1
+            x1 = x
+            y2 = y1
+            y1 = y
             buf[i] = y
-        self._dc_prev_x = px
-        self._dc_prev_y = py
+        self._notch_x1 = x1
+        self._notch_x2 = x2
+        self._notch_y1 = y1
+        self._notch_y2 = y2
         
         for idx, entry in enumerate(chain):
             if entry["bypass"] or not entry["enabled"]:
@@ -707,22 +721,27 @@ class AudioPipeline:
         ch0 = audio_in[0, :].copy()
         ch1 = audio_in[1, :].copy()
 
-        # ── DC-blocking high-pass filter on both channels ───────────
-        alpha = self._dc_alpha
-        px = self._dc_prev_x
-        py = self._dc_prev_y
+        # ── 60Hz mains hum notch filter on both channels ────────────
+        b0, b1, b2 = self._notch_b0, self._notch_b1, self._notch_b2
+        a1, a2 = self._notch_a1, self._notch_a2
+        # Apply notch to ch0
+        x1, x2 = self._notch_x1, self._notch_x2
+        y1, y2 = self._notch_y1, self._notch_y2
         for i in range(len(ch0)):
-            y = alpha * (py + ch0[i] - px)
-            px = ch0[i]
-            py = y
+            x = ch0[i]
+            y = b0*x + b1*x1 + b2*x2 - a1*y1 - a2*y2
+            x2 = x1; x1 = x; y2 = y1; y1 = y
             ch0[i] = y
+        # Apply notch to ch1 (continues from ch0's state)
         for i in range(len(ch1)):
-            y = alpha * (py + ch1[i] - px)
-            px = ch1[i]
-            py = y
+            x = ch1[i]
+            y = b0*x + b1*x1 + b2*x2 - a1*y1 - a2*y2
+            x2 = x1; x1 = x; y2 = y1; y1 = y
             ch1[i] = y
-        self._dc_prev_x = px
-        self._dc_prev_y = py
+        self._notch_x1 = x1
+        self._notch_x2 = x2
+        self._notch_y1 = y1
+        self._notch_y2 = y2
 
         # Update input VU level from the guitar input channel (ch0)
         in_rms = np.sqrt(np.mean(ch0 ** 2) + _EPS)
@@ -3143,7 +3162,13 @@ class AudioPipeline:
         self._block_size = block_size
         self._sample_rate = sample_rate
         self._vu_alpha = np.exp(-block_size / (0.05 * sample_rate))
-        self._dc_alpha = 1.0 / (1.0 + 2.0 * np.pi * 15.0 / sample_rate)
+        # Recompute 60Hz notch coefficients for new sample rate
+        _b0, _b1, _b2, _a1, _a2 = _compute_notch_coeffs(60.0, 5.0, sample_rate)
+        self._notch_b0, self._notch_b1, self._notch_b2 = _b0, _b1, _b2
+        self._notch_a1, self._notch_a2 = _a1, _a2
+        # Reset notch filter state to avoid pop on rate change
+        self._notch_x1 = self._notch_x2 = 0.0
+        self._notch_y1 = self._notch_y2 = 0.0
         # Clear DSP state — effects will reinit with new block/sample rate
         self._state.clear()
         self._coeffs.clear()