From f5b220388428a8161908bed824daa9f28dbd67ba Mon Sep 17 00:00:00 2001
From: liangming <liangming@liangmingdeMac-mini.local>
Date: Sun, 23 Aug 2026 14:37:12 +0800
Subject: [PATCH] Restore audio transcription mode configuration

---
 google/genai/tests/types/test_types.py |  8 +++++++-
 google/genai/types.py                  | 16 ++++++++++++++++
 2 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/google/genai/tests/types/test_types.py b/google/genai/tests/types/test_types.py
index 50a3f9058..eae8dffdf 100644
--- a/google/genai/tests/types/test_types.py
+++ b/google/genai/tests/types/test_types.py
@@ -46,6 +46,13 @@ class SubFunctionResponsePart(types.FunctionResponsePart):
   pass
 
 
+def test_audio_transcription_config_mode():
+  config = types.AudioTranscriptionConfig(mode='smart')
+
+  assert config.mode == types.AudioTranscriptionConfigMode.SMART
+  assert config.model_dump(by_alias=True, exclude_none=True)['mode'] == 'SMART'
+
+
 def test_factory_method_from_uri_part():
 
   my_part = SubPart.from_uri(
@@ -2969,4 +2976,3 @@ def test_computer_use_types():
   assert c.enable_prompt_injection_detection is True
   assert len(c.disabled_safety_policies) == 2
   assert types.SafetyPolicy.FINANCIAL_TRANSACTIONS in c.disabled_safety_policies
-
diff --git a/google/genai/types.py b/google/genai/types.py
index 0cbf8f5eb..9243603a7 100644
--- a/google/genai/types.py
+++ b/google/genai/types.py
@@ -6338,6 +6338,15 @@ class LanguageHintsDict(TypedDict, total=False):
 LanguageHintsOrDict = Union[LanguageHints, LanguageHintsDict]
 
 
+class AudioTranscriptionConfigMode(_common.CaseInSensitiveEnum):
+  """Controls the level of detail returned by audio transcription."""
+
+  VERBATIM = 'VERBATIM'
+  """Returns a verbatim transcription."""
+  SMART = 'SMART'
+  """Returns a cleaned-up transcription."""
+
+
 class AudioTranscriptionConfig(_common.BaseModel):
   """The audio transcription configuration in Setup."""
 
@@ -6371,6 +6380,10 @@ class AudioTranscriptionConfig(_common.BaseModel):
       description="""Configures speaker diarization.
       """,
   )
+  mode: Optional[AudioTranscriptionConfigMode] = Field(
+      default=None,
+      description="""Controls the level of detail returned by audio transcription.""",
+  )
 
 
 class AudioTranscriptionConfigDict(TypedDict, total=False):
@@ -6399,6 +6412,9 @@ class AudioTranscriptionConfigDict(TypedDict, total=False):
   """Configures speaker diarization.
       """
 
+  mode: Optional[AudioTranscriptionConfigMode]
+  """Controls the level of detail returned by audio transcription."""
+
 
 AudioTranscriptionConfigOrDict = Union[
     AudioTranscriptionConfig, AudioTranscriptionConfigDict