diff --git a/src/spikeinterface/core/analyzer_extension_core.py b/src/spikeinterface/core/analyzer_extension_core.py
index 7d3d890f7f..f1c2cb5a5b 100644
--- a/src/spikeinterface/core/analyzer_extension_core.py
+++ b/src/spikeinterface/core/analyzer_extension_core.py
@@ -156,7 +156,7 @@ def _get_waveforms(self, sorting_analyzer=None, unit_ids=None, verbose=False, **
 
         recording = self.sorting_analyzer.recording
         sorting = self.sorting_analyzer.sorting
-        
+
         some_spikes = sorting_analyzer.get_extension("random_spikes").get_random_spikes()
 
         if unit_ids is None:
@@ -246,27 +246,29 @@ def _merge_extension_data(self, units_to_merge, new_unit_ids, new_sorting_analyz
         if new_sorting_analyzer.sparsity is not None:
             sparsity_mask = new_sorting_analyzer.sparsity.mask
             num_chans = int(max(np.sum(sparsity_mask, axis=1)))
-            old_num_chans = self.data['waveforms'].shape[2]
+            old_num_chans = self.data["waveforms"].shape[2]
             if num_chans == old_num_chans:
-                new_data['waveforms'] = self.data['waveforms']
+                new_data["waveforms"] = self.data["waveforms"]
             else:
-                num_waveforms = len(self.data['waveforms'])
-                num_samples = self.data['waveforms'].shape[1]
+                num_waveforms = len(self.data["waveforms"])
+                num_samples = self.data["waveforms"].shape[1]
 
                 some_spikes = new_sorting_analyzer.get_extension("random_spikes").get_random_spikes()
-                new_data['waveforms'] = np.zeros((num_waveforms, num_samples, num_chans), dtype=self.data['waveforms'].dtype)
+                new_data["waveforms"] = np.zeros(
+                    (num_waveforms, num_samples, num_chans), dtype=self.data["waveforms"].dtype
+                )
                 keep_unit_indices = np.flatnonzero(~np.isin(new_sorting_analyzer.unit_ids, new_unit_ids))
                 keep_spike_mask = np.isin(some_spikes["unit_index"], keep_unit_indices)
-                new_data['waveforms'][keep_spike_mask, :, :old_num_chans] = self.data['waveforms'][keep_spike_mask]
+                new_data["waveforms"][keep_spike_mask, :, :old_num_chans] = self.data["waveforms"][keep_spike_mask]
 
-                # We only recompute waveforms for new units that might have a new sparsity mask. Could be 
+                # We only recompute waveforms for new units that might have a new sparsity mask. Could be
                 # slightly optimized by checking exactly which merged units have a different mask
                 updated_unit_indices = np.flatnonzero(np.isin(new_sorting_analyzer.unit_ids, new_unit_ids))
                 updated_spike_mask = np.isin(some_spikes["unit_index"], updated_unit_indices)
                 new_waveforms = self._get_waveforms(new_sorting_analyzer, new_unit_ids, verbose, **job_kwargs)
-                new_data['waveforms'][updated_spike_mask] = new_waveforms
+                new_data["waveforms"][updated_spike_mask] = new_waveforms
         else:
-            new_data['waveforms'] = self.data['waveforms']
+            new_data["waveforms"] = self.data["waveforms"]
 
         return new_data
 
@@ -495,7 +497,6 @@ def _merge_extension_data(self, units_to_merge, new_unit_ids, new_sorting_analyz
 
         return new_data
 
-
     def _get_data(self, operator="average", percentile=None, outputs="numpy"):
         if operator != "percentile":
             key = operator