diff --git a/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field.py b/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field.py
index 056899b6e..d5a1d260e 100644
--- a/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field.py
+++ b/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field.py
@@ -96,12 +96,14 @@ def _cast_value(ctx: Context, value: Any, data_type: type | term.URIRef | None):
         else:
             raise ValueError(f"Type {type(value)} is not accepted for an image.")
     elif data_type == DataType.AUDIO_OBJECT:
-        output = deps.librosa.load(io.BytesIO(value))
+        output = deps.librosa.load(io.BytesIO(value), sr=None)
         return output
     elif data_type == DataType.BOUNDING_BOX:  # pytype: disable=wrong-arg-types
         return bounding_box.parse(value)
     elif not isinstance(data_type, type):
         raise ValueError(f"No special case for type {data_type}.")
+    elif isinstance(value, np.ndarray) and issubclass(data_type, np.generic):
+        return value.astype(data_type)
     elif isinstance(value, list) or isinstance(value, np.ndarray):
         return [_cast_value(ctx=ctx, value=v, data_type=data_type) for v in value]
     elif data_type == bytes and not isinstance(value, bytes):
diff --git a/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field_test.py b/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field_test.py
index 4299193b6..f71d8c3b1 100644
--- a/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field_test.py
+++ b/python/mlcroissant/mlcroissant/_src/operation_graph/operations/field_test.py
@@ -51,6 +51,21 @@ def test_cast_value(conforms_to, value, data_type, expected):
     assert field._cast_value(ctx, value, data_type) == expected
 
 
+@parametrize_conforms_to()
+@pytest.mark.parametrize(
+    ["value", "data_type", "expected"],
+    [
+        [np.array([1, 2, 3]), DataType.INTEGER, np.array([1, 2, 3])],
+        [np.array([1, 2, 3]), DataType.FLOAT32, np.array([1.0, 2.0, 3.0])],
+    ],
+)
+def test_cast_value_ndarray():
+    ctx = Context(conforms_to=conforms_to)
+    cast_value = field._cast_value(ctx, value, data_type)
+    assert cast_value == expected
+    assert cast_value.dtype == expected.dtype
+
+
 @parametrize_conforms_to()
 @pytest.mark.parametrize(
     ["value", "data_type"],