diff --git a/assemblyai/_audio.py b/assemblyai/_audio.py
index e0bab83..3756ac9 100644
--- a/assemblyai/_audio.py
+++ b/assemblyai/_audio.py
@@ -90,7 +90,9 @@ def _resolve_audio(
     elif hasattr(data, "read"):
         audio = data.read()
         name = getattr(data, "name", None)
-        if name:
+        # `name` is a file descriptor (an int) for anonymous files such as
+        # `tempfile.TemporaryFile()`, so only a str is a usable path.
+        if isinstance(name, str) and name:
             filename = os.path.basename(name)
             suffix = os.path.splitext(name)[1].lower()
     else:
diff --git a/tests/unit/test_sync.py b/tests/unit/test_sync.py
index 054101c..9aac646 100644
--- a/tests/unit/test_sync.py
+++ b/tests/unit/test_sync.py
@@ -1,3 +1,5 @@
+import tempfile
+
 import httpx
 import pytest
 from pytest_httpx import HTTPXMock
@@ -316,6 +318,24 @@ def test_transcribe_path_input(httpx_mock: HTTPXMock, tmp_path):
     assert b'filename="call.wav"' in body
 
 
+def test_transcribe_file_object_without_a_path_name(httpx_mock: HTTPXMock):
+    # Given an anonymous temporary file, whose `name` is a file descriptor
+    # rather than a path
+    _mock_ok(httpx_mock)
+    with tempfile.TemporaryFile() as audio_file:
+        audio_file.write(b"RIFFfake-wav-bytes")
+        audio_file.seek(0)
+
+        # When transcribing the file object
+        result = aai.SyncTranscriber().transcribe(audio_file)
+
+    # Then it succeeds and ships the audio under the default name
+    assert result.text == "hello world"
+    body = httpx_mock.get_requests()[0].read()
+    assert b'filename="audio.wav"' in body
+    assert b"RIFFfake-wav-bytes" in body
+
+
 def test_keyterms_prompt_too_long_raises():
     # Given a keyterms_prompt exceeding the character cap
     # When building the config, Then validation fails immediately