diff --git a/shared-bindings/ulab/filter/__init__.pyi b/shared-bindings/ulab/filter/__init__.pyi
index fff404300b..5e7202e06c 100644
--- a/shared-bindings/ulab/filter/__init__.pyi
+++ b/shared-bindings/ulab/filter/__init__.pyi
@@ -17,3 +17,20 @@ def convolve(r, c=None):
 
     Convolution is most time-efficient when both inputs are of float type."""
     ...
+
+def sosfilt(sos : ulab.array, x : ulab.array, *, xi : Optional[ulab.array] = None) -> Union[ulab.array, Tuple[ulab.array, ulab.array]]:
+    """
+    :param ulab.array sos: Array of second-order filter coefficients, must have shape (n_sections, 6). Each row corresponds to a second-order section, with the first three columns providing the numerator coefficients and the last three providing the denominator coefficients.
+    :param ulab.array x: The data to be filtered
+    :param ulab.array zi: Optional initial conditions for the filter
+
+    :return: If ``xi`` is not specified, the filter result alone is returned.  If ``xi`` is specified, the return value is a 2-tuple of the filter result and the final filter conditions.
+
+    Filter data along one dimension using cascaded second-order sections.
+
+    Filter a data sequence, x, using a digital IIR filter defined by sos.
+
+    The filter function is implemented as a series of second-order filters with direct-form II transposed structure. It is designed to minimize numerical precision errors for high-order filters.
+
+    Filter coefficients can be generated by using scipy's filter generators such as ``signal.ellip(..., output='sos')``."""
+    ...