feat(azure/gpt_transformation.py): add azure audio model support

Closes https://github.com/BerriAI/litellm/issues/6305

litellm/llms/azure/chat/gpt_transformation.py

@@ -99,6 +99,8 @@ class AzureOpenAIConfig(BaseConfig):
             "extra_headers",
             "parallel_tool_calls",
             "prediction",
+            "modalities",
+            "audio",
         ]
 
     def _is_response_format_supported_model(self, model: str) -> bool:

litellm/llms/openai/chat/gpt_transformation.py

@@ -125,6 +125,7 @@ class OpenAIGPTConfig(BaseLLMModelInfo, BaseConfig):
             "max_retries",
             "extra_headers",
             "parallel_tool_calls",
+            "audio",
         ]  # works across all models
 
         model_specific_params = []

tests/llm_translation/test_optional_params.py

@@ -1368,3 +1368,14 @@ def test_bedrock_invoke_anthropic_max_tokens():
 
     assert "max_tokens_to_sample" not in optional_params
     assert optional_params["max_tokens"] == 1024
+
+
+def test_azure_modalities_param():
+    optional_params = get_optional_params(
+        model="chatgpt-v2",
+        custom_llm_provider="azure",
+        modalities=["text", "audio"],
+        audio={"type": "audio_input", "input": "test.wav"},
+    )
+    assert optional_params["modalities"] == ["text", "audio"]
+    assert optional_params["audio"] == {"type": "audio_input", "input": "test.wav"}