diff --git a/litellm/tests/eagle.wav b/litellm/tests/eagle.wav
new file mode 100644
index 000000000..1c2365785
Binary files /dev/null and b/litellm/tests/eagle.wav differ
diff --git a/litellm/tests/test_whisper.py b/litellm/tests/test_whisper.py
index 9d26d2d4e..0ac8d094f 100644
--- a/litellm/tests/test_whisper.py
+++ b/litellm/tests/test_whisper.py
@@ -26,6 +26,10 @@ file_path = os.path.join(pwd, "gettysburg.wav")
 
 audio_file = open(file_path, "rb")
 
+
+file2_path = os.path.join(pwd, "eagle.wav")
+audio_file2 = open(file2_path, "rb")
+
 load_dotenv()
 
 sys.path.insert(
@@ -148,3 +152,44 @@ async def test_transcription_on_router():
     except Exception as e:
         traceback.print_exc()
         pytest.fail(f"Error occurred: {e}")
+
+
+@pytest.mark.asyncio()
+async def test_transcription_caching():
+    import litellm
+    from litellm.caching import Cache
+
+    litellm.set_verbose = True
+    litellm.cache = Cache()
+
+    # make raw llm api call
+
+    response_1 = await litellm.atranscription(
+        model="whisper-1",
+        file=audio_file,
+    )
+
+    await asyncio.sleep(5)
+
+    # cache hit
+
+    response_2 = await litellm.atranscription(
+        model="whisper-1",
+        file=audio_file,
+    )
+
+    print("response_1", response_1)
+    print("response_2", response_2)
+    print("response2 hidden params", response_2._hidden_params)
+    assert response_2._hidden_params["cache_hit"] is True
+
+    # cache miss
+
+    response_3 = await litellm.atranscription(
+        model="whisper-1",
+        file=audio_file2,
+    )
+    print("response_3", response_3)
+    print("response3 hidden params", response_3._hidden_params)
+    assert response_3._hidden_params.get("cache_hit") is not True
+    assert response_3.text != response_2.text