diff --git a/litellm/litellm_core_utils/get_supported_openai_params.py b/litellm/litellm_core_utils/get_supported_openai_params.py index 3020718cc30..0fe70138361 100644 --- a/litellm/litellm_core_utils/get_supported_openai_params.py +++ b/litellm/litellm_core_utils/get_supported_openai_params.py @@ -122,10 +122,9 @@ def get_supported_openai_params( transcription_provider_config = litellm.ProviderConfigManager.get_provider_audio_transcription_config( model=model, provider=LlmProviders.OPENAI ) - if isinstance(transcription_provider_config, litellm.OpenAIGPTAudioTranscriptionConfig): + if transcription_provider_config is not None: return transcription_provider_config.get_supported_openai_params(model=model) - else: - raise ValueError(f"Unsupported provider config: {transcription_provider_config} for model: {model}") + raise ValueError(f"Unsupported provider config: {transcription_provider_config} for model: {model}") return litellm.OpenAIConfig().get_supported_openai_params(model=model) elif custom_llm_provider == "sap": if request_type == "chat_completion": diff --git a/tests/litellm_utils_tests/test_utils.py b/tests/litellm_utils_tests/test_utils.py index 2017e603af1..f3016f34df1 100644 --- a/tests/litellm_utils_tests/test_utils.py +++ b/tests/litellm_utils_tests/test_utils.py @@ -2148,6 +2148,20 @@ def test_gemini_model_info_reports_transcription_params(monkeypatch): assert "keywords" in model_info["supported_openai_params"] +def test_gemini_live_transcription_dispatch_has_no_batch_params(): + from litellm.litellm_core_utils.get_supported_openai_params import ( + get_supported_openai_params, + ) + + params = get_supported_openai_params( + model="gemini-3.5-transcribe-live", + custom_llm_provider="gemini", + request_type="transcription", + ) + + assert params is None + + @pytest.mark.parametrize( "model, expected_bool", [ diff --git a/tests/test_litellm/litellm_core_utils/test_audio_utils.py b/tests/test_litellm/litellm_core_utils/test_audio_utils.py index 155f6680416..146d0c38a06 100644 --- a/tests/test_litellm/litellm_core_utils/test_audio_utils.py +++ b/tests/test_litellm/litellm_core_utils/test_audio_utils.py @@ -15,6 +15,7 @@ from litellm.litellm_core_utils.audio_utils.utils import ( get_audio_file_content_hash, get_audio_file_for_health_check, get_audio_file_name, + normalize_transcription_keywords, process_audio_file, ) @@ -349,6 +350,19 @@ class TestNormalizeTranscriptionLanguageToBcp47: assert normalize_transcription_language_to_bcp47(language) == expected +class TestNormalizeTranscriptionKeywords: + @pytest.mark.parametrize( + ("keywords", "expected"), + [ + (["alpha", "", 42, "beta"], ("alpha", "beta")), + ("alpha", ()), + (None, ()), + ], + ) + def test_normalization(self, keywords, expected): + assert normalize_transcription_keywords(keywords) == expected + + class TestResolveSpeechMediaType: @pytest.mark.parametrize( ("upstream_content_type", "response_format", "expected"), diff --git a/tests/unit/llms/gemini/realtime/test_gemini_realtime_transformation.py b/tests/unit/llms/gemini/realtime/test_gemini_realtime_transformation.py index fde06dfbc50..a75947d138e 100644 --- a/tests/unit/llms/gemini/realtime/test_gemini_realtime_transformation.py +++ b/tests/unit/llms/gemini/realtime/test_gemini_realtime_transformation.py @@ -864,6 +864,38 @@ def test_gemini_session_update_maps_input_audio_transcription_settings(): } +@pytest.mark.parametrize( + ("transcription", "expected"), + [ + ("invalid", {}), + ({}, {}), + ({"language": "", "keywords": ["", 42]}, {}), + ( + {"language": "pl", "keywords": ["", 42, "term"]}, + { + "languageCodes": ["pl-PL"], + "customVocabulary": ["term"], + }, + ), + ], +) +def test_gemini_session_update_handles_input_audio_transcription_values(transcription, expected): + config = GeminiRealtimeConfig() + session_update = { + "type": "session.update", + "session": {"input_audio_transcription": transcription}, + } + + result = config.transform_realtime_request( + json.dumps(session_update), + "gemini-3.5-transcribe-live", + session_configuration_request=None, + ) + + setup = json.loads(result[0]) + assert setup["setup"]["inputAudioTranscription"] == expected + + def test_gemini_session_update_includes_input_audio_transcription_default(): """Verify _handle_session_update includes inputAudioTranscription default.""" config = GeminiRealtimeConfig()