diff --git a/readalongs/text/util.py b/readalongs/text/util.py
index 6243f4f6..68cf2980 100644
--- a/readalongs/text/util.py
+++ b/readalongs/text/util.py
@@ -14,6 +14,7 @@
 from collections import OrderedDict
 from datetime import datetime
 from io import TextIOWrapper
+from pathlib import Path
 from typing import IO, Union
 from unicodedata import normalize
 
@@ -104,7 +105,7 @@ def is_do_not_align(element):
     return dna in ("true", "True", "TRUE", "1")
 
 
-def load_xml(input_path: Union[str, IO]) -> etree.ElementTree:
+def load_xml(input_path: Union[str, Path, IO]) -> etree.ElementTree:
     """Safely load an XML file with etree.parse to respect encoding
 
     Return: the root of the XML etree
diff --git a/test/test_api.py b/test/test_api.py
index 52374b3f..0619f2e9 100755
--- a/test/test_api.py
+++ b/test/test_api.py
@@ -34,13 +34,13 @@ def test_call_align(self):
         self.assertTrue(exception is None)
         self.assertIn("Words (<w>) not present; tokenizing", log)
         expected_output_files = (
-            "output.readalong",
-            "output.m4a",
+            "www/output.readalong",
+            "www/output.m4a",
             "output.TextGrid",
             "output_sentences.srt",
             "output_words.srt",
-            "index.html",
-            "output.html",
+            "www/index.html",
+            "Offline-HTML/output.html",
         )
         for f in expected_output_files:
             self.assertTrue(
diff --git a/test/test_audio.py b/test/test_audio.py
index 675af5eb..5814c413 100755
--- a/test/test_audio.py
+++ b/test/test_audio.py
@@ -81,7 +81,7 @@ def test_align_sample(self):
         if process.returncode != 0:
             LOGGER.error("Subprocess readalongs align failed: %s", process.stderr)
         # Check Result
-        raspath = Path(output_path)
+        raspath = Path(output_path) / "www"
         ras_files = raspath.glob("*.readalong")
         self.assertTrue(
             next(ras_files, False),
@@ -108,7 +108,7 @@ def test_align_removed(self):
         if process.returncode != 0:
             LOGGER.error("Subprocess readalongs align failed: %s", process.stderr)
         # Check Result
-        raspath = Path(output_path)
+        raspath = Path(output_path) / "www"
         ras_files = raspath.glob("*.readalong")
         self.assertTrue(
             next(ras_files, False),
@@ -135,7 +135,7 @@ def test_align_muted(self):
         if process.returncode != 0:
             LOGGER.error("Subprocess readalongs align failed: %s", process.stderr)
         # Check Result
-        raspath = Path(output_path)
+        raspath = Path(output_path) / "www"
         ras_files = raspath.glob("*.readalong")
         self.assertTrue(
             next(ras_files, False),
diff --git a/test/test_silence.py b/test/test_silence.py
index 69d52b75..fb23d81a 100755
--- a/test/test_silence.py
+++ b/test/test_silence.py
@@ -17,7 +17,7 @@ class TestSilence(BasicTestCase):
 
     def test_basic_silence_insertion(self):
         """Basic usage of the silence feature in a readalong"""
-        output = os.path.join(self.tempdir, "silence")
+        output = self.tempdir / "silence"
         # Run align from xml
         results = self.runner.invoke(
             align,
@@ -35,22 +35,20 @@ def test_basic_silence_insertion(self):
                 "fra",
                 os.path.join(self.data_dir, "ej-fra-silence.readalong"),
                 os.path.join(self.data_dir, "ej-fra.m4a"),
-                output,
+                str(output),
             ],
         )
         self.assertEqual(results.exit_code, 0)
-        self.assertTrue(os.path.exists(os.path.join(output, "silence.m4a")))
+        self.assertTrue((output / "www/silence.m4a").exists())
         # test silence spans in output xml
-        root = load_xml(os.path.join(output, "silence.readalong"))
+        root = load_xml(output / "www/silence.readalong")
         silence_spans = root.xpath("//silence")
         self.assertEqual(len(silence_spans), 3)
         # test audio has correct amount of silence added
         original_audio = AudioSegment.from_file(
             os.path.join(self.data_dir, "ej-fra.m4a")
         )
-        new_audio = AudioSegment.from_file(
-            os.path.join(output, "silence.m4a"), format="m4a"
-        )
+        new_audio = AudioSegment.from_file(output / "www/silence.m4a", format="m4a")
         self.assertAlmostEqual(
             len(new_audio) - len(original_audio),
             2882,
@@ -59,7 +57,7 @@ def test_basic_silence_insertion(self):
         )
 
     def test_bad_silence(self):
-        output = os.path.join(self.tempdir, "bad_silence")
+        output = self.tempdir / "bad_silence"
         # Run align from bad xml
         results = self.runner.invoke(
             align,
@@ -77,7 +75,7 @@ def test_bad_silence(self):
                 "fra",
                 os.path.join(self.data_dir, "ej-fra-silence-bad.readalong"),
                 os.path.join(self.data_dir, "ej-fra.m4a"),
-                output,
+                str(output),
             ],
         )
         self.assertNotEqual(results.exit_code, 0)