[Bug Fix] Timestamp Granularities are not properly passed to whisper in Azure (#10299)

* test fix form data parsing

* test fix form data parsing

* fix types

litellm/proxy/common_utils/http_parsing_utils.py

@@ -1,5 +1,5 @@
 import json
-from typing import Dict, List, Optional
+from typing import Any, Dict, List, Optional
 
 import orjson
 from fastapi import Request, UploadFile, status
@@ -147,10 +147,10 @@ def check_file_size_under_limit(
 
     if llm_router is not None and request_data["model"] in router_model_names:
         try:
-            deployment: Optional[
-                Deployment
-            ] = llm_router.get_deployment_by_model_group_name(
-                model_group_name=request_data["model"]
+            deployment: Optional[Deployment] = (
+                llm_router.get_deployment_by_model_group_name(
+                    model_group_name=request_data["model"]
+                )
             )
             if (
                 deployment
@@ -185,3 +185,23 @@ def check_file_size_under_limit(
             )
 
     return True
+
+
+async def get_form_data(request: Request) -> Dict[str, Any]:
+    """
+    Read form data from request
+
+    Handles when OpenAI SDKs pass form keys as `timestamp_granularities[]="word"` instead of `timestamp_granularities=["word", "sentence"]`
+    """
+    form = await request.form()
+    form_data = dict(form)
+    parsed_form_data: dict[str, Any] = {}
+    for key, value in form_data.items():
+
+        # OpenAI SDKs pass form keys as `timestamp_granularities[]="word"` instead of `timestamp_granularities=["word", "sentence"]`
+        if key.endswith("[]"):
+            clean_key = key[:-2]
+            parsed_form_data.setdefault(clean_key, []).append(value)
+        else:
+            parsed_form_data[key] = value
+    return parsed_form_data

litellm/proxy/proxy_server.py

@@ -179,6 +179,7 @@ from litellm.proxy.common_utils.html_forms.ui_login import html_form
 from litellm.proxy.common_utils.http_parsing_utils import (
     _read_request_body,
     check_file_size_under_limit,
+    get_form_data,
 )
 from litellm.proxy.common_utils.load_config_utils import (
     get_config_file_contents_from_gcs,
@@ -804,9 +805,9 @@ model_max_budget_limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(
     dual_cache=user_api_key_cache
 )
 litellm.logging_callback_manager.add_litellm_callback(model_max_budget_limiter)
-redis_usage_cache: Optional[
-    RedisCache
-] = None  # redis cache used for tracking spend, tpm/rpm limits
+redis_usage_cache: Optional[RedisCache] = (
+    None  # redis cache used for tracking spend, tpm/rpm limits
+)
 user_custom_auth = None
 user_custom_key_generate = None
 user_custom_sso = None
@@ -1132,9 +1133,9 @@ async def update_cache(  # noqa: PLR0915
         _id = "team_id:{}".format(team_id)
         try:
             # Fetch the existing cost for the given user
-            existing_spend_obj: Optional[
-                LiteLLM_TeamTable
-            ] = await user_api_key_cache.async_get_cache(key=_id)
+            existing_spend_obj: Optional[LiteLLM_TeamTable] = (
+                await user_api_key_cache.async_get_cache(key=_id)
+            )
             if existing_spend_obj is None:
                 # do nothing if team not in api key cache
                 return
@@ -2806,9 +2807,9 @@ async def initialize(  # noqa: PLR0915
         user_api_base = api_base
         dynamic_config[user_model]["api_base"] = api_base
     if api_version:
-        os.environ[
-            "AZURE_API_VERSION"
-        ] = api_version  # set this for azure - litellm can read this from the env
+        os.environ["AZURE_API_VERSION"] = (
+            api_version  # set this for azure - litellm can read this from the env
+        )
     if max_tokens:  # model-specific param
         dynamic_config[user_model]["max_tokens"] = max_tokens
     if temperature:  # model-specific param
@@ -4120,7 +4121,7 @@ async def audio_transcriptions(
     data: Dict = {}
     try:
         # Use orjson to parse JSON data, orjson speeds up requests significantly
-        form_data = await request.form()
+        form_data = await get_form_data(request)
         data = {key: value for key, value in form_data.items() if key != "file"}
 
         # Include original request and headers in the data
@@ -7758,9 +7759,9 @@ async def get_config_list(
                             hasattr(sub_field_info, "description")
                             and sub_field_info.description is not None
                         ):
-                            nested_fields[
-                                idx
-                            ].field_description = sub_field_info.description
+                            nested_fields[idx].field_description = (
+                                sub_field_info.description
+                            )
                         idx += 1
 
                     _stored_in_db = None

tests/litellm/proxy/common_utils/test_http_parsing_utils.py

@@ -18,6 +18,7 @@ from litellm.proxy.common_utils.http_parsing_utils import (
     _read_request_body,
     _safe_get_request_parsed_body,
     _safe_set_request_parsed_body,
+    get_form_data,
 )
 
 
@@ -147,3 +148,53 @@ async def test_circular_reference_handling():
     assert (
         "proxy_server_request" not in result2
     )  # This will pass, showing the cache pollution
+
+
+@pytest.mark.asyncio
+async def test_get_form_data():
+    """
+    Test that get_form_data correctly handles form data with array notation.
+    Tests audio transcription parameters as a specific example.
+    """
+    # Create a mock request with transcription form data
+    mock_request = MagicMock()
+
+    # Create mock form data with array notation for timestamp_granularities
+    mock_form_data = {
+        "file": "file_object",  # In a real request this would be an UploadFile
+        "model": "gpt-4o-transcribe",
+        "include[]": "logprobs",  # Array notation
+        "language": "en",
+        "prompt": "Transcribe this audio file",
+        "response_format": "json",
+        "stream": "false",
+        "temperature": "0.2",
+        "timestamp_granularities[]": "word",  # First array item
+        "timestamp_granularities[]": "segment",  # Second array item (would overwrite in dict, but handled by the function)
+    }
+
+    # Mock the form method to return the test data
+    mock_request.form = AsyncMock(return_value=mock_form_data)
+
+    # Call the function being tested
+    result = await get_form_data(mock_request)
+
+    # Verify regular form fields are preserved
+    assert result["file"] == "file_object"
+    assert result["model"] == "gpt-4o-transcribe"
+    assert result["language"] == "en"
+    assert result["prompt"] == "Transcribe this audio file"
+    assert result["response_format"] == "json"
+    assert result["stream"] == "false"
+    assert result["temperature"] == "0.2"
+
+    # Verify array fields are correctly parsed
+    assert "include" in result
+    assert isinstance(result["include"], list)
+    assert "logprobs" in result["include"]
+
+    assert "timestamp_granularities" in result
+    assert isinstance(result["timestamp_granularities"], list)
+    # Note: In a real MultiDict, both values would be present
+    # But in our mock dictionary the second value overwrites the first
+    assert "segment" in result["timestamp_granularities"]