diff --git a/nodes/selva_bigvgan_trainer.py b/nodes/selva_bigvgan_trainer.py
index e27bfd6..f235d17 100644
--- a/nodes/selva_bigvgan_trainer.py
+++ b/nodes/selva_bigvgan_trainer.py
@@ -241,19 +241,17 @@ class SelvaBigvganTrainer:
                                     fresh, requires_grad=True
                                 )
 
-                    # mel_converter buffers (mel_basis, hann_window, etc.) were loaded
-                    # inside ComfyUI's outer inference_mode context, so they are inference
-                    # tensors. Operations on inference tensors ALWAYS produce inference
-                    # tensors, even inside inference_mode(False). torch.zeros() et al.
-                    # create normal tensors in the current (non-inference) context, so
-                    # we replace every buffer once via copy_() to break the chain.
-                    for bname, buf in list(mel_converter._buffers.items()):
-                        if buf is not None:
-                            fresh = torch.zeros(
-                                buf.shape, device=buf.device, dtype=buf.dtype
-                            )
-                            fresh.copy_(buf)
-                            mel_converter._buffers[bname] = fresh
+                    # mel_converter and its submodules (e.g. Spectrogram.window) have
+                    # inference-tensor buffers loaded in ComfyUI's outer inference_mode.
+                    # Must iterate .modules() — ._buffers only covers direct buffers.
+                    for sub in mel_converter.modules():
+                        for bname, buf in list(sub._buffers.items()):
+                            if buf is not None:
+                                fresh = torch.zeros(
+                                    buf.shape, device=buf.device, dtype=buf.dtype
+                                )
+                                fresh.copy_(buf)
+                                sub._buffers[bname] = fresh
 
                     optimizer = torch.optim.AdamW(
                         vocoder.parameters(), lr=lr, betas=(0.8, 0.99)
@@ -280,11 +278,16 @@ class SelvaBigvganTrainer:
                         del _stacked
                         target_wav = target_flat.unsqueeze(1)                 # [B, 1, T]
 
-                        # Fixed target mel — buffers are now normal tensors (sanitized
-                        # above), so torch.no_grad() correctly produces a non-inference,
-                        # no-grad leaf tensor that conv layers can save for backward.
+                        # Compute target mel and guarantee it is not an inference tensor.
+                        # Even with sanitized buffers a submodule we missed could still
+                        # taint the output, so we always copy into a fresh tensor.
                         with torch.no_grad():
-                            target_mel = mel_converter(target_flat)           # [B, 80, T_mel]
+                            _mel = mel_converter(target_flat)
+                        target_mel = torch.empty(
+                            _mel.shape, device=device, dtype=dtype
+                        )
+                        target_mel.copy_(_mel)
+                        del _mel
 
                         # Vocoder forward: mel → waveform
                         pred_wav = vocoder(target_mel)                        # [B, 1, T_wav]