From 526d5439f3fc57b30d6d5c084c0f113704827722 Mon Sep 17 00:00:00 2001
From: Abhishek Kumar <abhishek@a6k.me>
Date: Tue, 31 Mar 2026 18:17:44 +0530
Subject: [PATCH] feat: add language option

---
 api/services/configuration/registry.py  | 12 +++++++++++-
 api/services/pipecat/service_factory.py | 14 +++++++++-----
 2 files changed, 20 insertions(+), 6 deletions(-)

diff --git a/api/services/configuration/registry.py b/api/services/configuration/registry.py
index 6ec1e632..51939b77 100644
--- a/api/services/configuration/registry.py
+++ b/api/services/configuration/registry.py
@@ -318,6 +318,9 @@ OPENAI_REALTIME_VOICES = [
 
 GOOGLE_REALTIME_MODELS = ["gemini-3.1-flash-live-preview"]
 GOOGLE_REALTIME_VOICES = ["Puck", "Charon", "Kore", "Fenrir", "Aoede"]
+GOOGLE_REALTIME_LANGUAGES = [
+    "en"
+]
 
 
 @register_service(ServiceType.REALTIME)
@@ -326,7 +329,7 @@ class GoogleRealtimeLLMConfiguration(BaseLLMConfiguration):
         ServiceProviders.GOOGLE_REALTIME
     )
     model: str = Field(
-        default="gemini-2.0-flash-live-001",
+        default="gemini-3.1-flash-live-preview",
         json_schema_extra={
             "examples": GOOGLE_REALTIME_MODELS,
             "allow_custom_input": True,
@@ -339,6 +342,13 @@ class GoogleRealtimeLLMConfiguration(BaseLLMConfiguration):
             "allow_custom_input": True,
         },
     )
+    language: str = Field(
+        default="en",
+        json_schema_extra={
+            "examples": GOOGLE_REALTIME_LANGUAGES,
+            "allow_custom_input": True,
+        },
+    )
 
 
 REALTIME_PROVIDERS = {
diff --git a/api/services/pipecat/service_factory.py b/api/services/pipecat/service_factory.py
index 128ae67e..965a359d 100644
--- a/api/services/pipecat/service_factory.py
+++ b/api/services/pipecat/service_factory.py
@@ -415,9 +415,10 @@ def create_realtime_llm_service(user_config, audio_config: "AudioConfig"):
     model = realtime_config.model
     api_key = realtime_config.api_key
     voice = getattr(realtime_config, "voice", None)
+    language = getattr(realtime_config, "language", None)
 
     logger.info(
-        f"Creating realtime LLM service: provider={provider}, model={model}, voice={voice}"
+        f"Creating realtime LLM service: provider={provider}, model={model}, voice={voice}, language={language}"
     )
 
     if provider == ServiceProviders.OPENAI_REALTIME.value:
@@ -451,12 +452,15 @@ def create_realtime_llm_service(user_config, audio_config: "AudioConfig"):
 
         # Gemini Live enables input/output audio transcription by default
         # in its _connect() method — no need to configure it explicitly.
+        settings_kwargs = {
+            "model": model,
+            "voice": voice or "Puck",
+        }
+        if language:
+            settings_kwargs["language"] = language
         return GeminiLiveLLMService(
             api_key=api_key,
-            settings=GeminiLiveLLMService.Settings(
-                model=model,
-                voice=voice or "Puck",  # vad=GeminiVADParams(disabled=True)
-            ),
+            settings=GeminiLiveLLMService.Settings(**settings_kwargs),
         )
     else:
         raise HTTPException(