diff --git a/src/spikeinterface/core/numpyextractors.py b/src/spikeinterface/core/numpyextractors.py
index 7abb5596e8..9a04eeec37 100644
--- a/src/spikeinterface/core/numpyextractors.py
+++ b/src/spikeinterface/core/numpyextractors.py
@@ -302,7 +302,7 @@ def from_neo_spiketrain_list(neo_spiketrains, sampling_frequency, unit_ids=None)
         return sorting
 
     @staticmethod
-    def from_peaks(peaks, sampling_frequency, unit_ids=None) -> "NumpySorting":
+    def from_peaks(peaks, sampling_frequency, unit_ids) -> "NumpySorting":
         """
         Construct a sorting from peaks returned by 'detect_peaks()' function.
         The unit ids correspond to the recording channel ids and spike trains are the
@@ -314,6 +314,8 @@ def from_peaks(peaks, sampling_frequency, unit_ids=None) -> "NumpySorting":
             Peaks array as returned by the 'detect_peaks()' function
         sampling_frequency : float
             the sampling frequency in Hz
+        unit_ids: np.array
+            The unit_ids vector which is generally the channel_ids but can be different.
 
         Returns
         -------
@@ -325,9 +327,6 @@ def from_peaks(peaks, sampling_frequency, unit_ids=None) -> "NumpySorting":
         spikes["unit_index"] = peaks["channel_index"]
         spikes["segment_index"] = peaks["segment_index"]
 
-        if unit_ids is None:
-            unit_ids = np.unique(peaks["channel_index"])
-
         sorting = NumpySorting(spikes, sampling_frequency, unit_ids)
 
         return sorting
diff --git a/src/spikeinterface/sortingcomponents/waveforms/temporal_pca.py b/src/spikeinterface/sortingcomponents/waveforms/temporal_pca.py
index a9e6126ccc..95a5166d10 100644
--- a/src/spikeinterface/sortingcomponents/waveforms/temporal_pca.py
+++ b/src/spikeinterface/sortingcomponents/waveforms/temporal_pca.py
@@ -135,7 +135,7 @@ def fit(
         peaks = select_peaks(peaks, **peak_selection_params)  # How to select n_peaks
 
         # Creates a numpy sorting object where the spike times are the peak times and the unit ids are the peak channel
-        sorting = NumpySorting.from_peaks(peaks, sampling_frequency=recording.sampling_frequency)
+        sorting = NumpySorting.from_peaks(peaks, recording.sampling_frequency, recording.channel_ids)
         # Create a waveform extractor
         we = extract_waveforms(
             recording,