Update app.py

app.py

@@ -17,7 +17,6 @@ from fastrtc import (
     wait_for_item,
 )
 from google import genai
-from google.genai import types
 from google.genai.types import (
     LiveConnectConfig,
     PrebuiltVoiceConfig,
@@ -30,9 +29,12 @@ from google.genai.types import (
 )
 from gradio.utils import get_space
 from pydantic import BaseModel
+from langdetect import detect, DetectorFactory
 
-current_dir = pathlib.Path(__file__).parent
+# Make langdetect results consistent
+DetectorFactory.seed = 0
 
+current_dir = pathlib.Path(__file__).parent
 load_dotenv()
 
 
@@ -41,6 +43,20 @@ def encode_audio(data: np.ndarray) -> str:
     return base64.b64encode(data.tobytes()).decode("UTF-8")
 
 
+def detect_language_code(text: str) -> str:
+    """Detect if the text is in English or Spanish and return Google TTS code"""
+    try:
+        lang = detect(text)
+        if lang.startswith("es"):
+            return "es-ES"  # Spanish
+        elif lang.startswith("en"):
+            return "en-US"  # English
+        else:
+            return "en-US"  # default
+    except Exception:
+        return "en-US"
+
+
 class GeminiHandler(AsyncStreamHandler):
     """Handler for the Gemini API"""
 
@@ -69,26 +85,26 @@ class GeminiHandler(AsyncStreamHandler):
             await self.wait_for_args()
             api_key, voice_name, system_message = self.latest_args[1:]
         else:
-            api_key, voice_name, system_message = None, "Kore", "Du bist ein hilfsamer Assistent."
+            api_key, voice_name, system_message = None, "Kore", "You are a helpful assistant."
+
+        # Auto-detect language from system_message
+        lang_code = detect_language_code(system_message)
 
         client = genai.Client(
             api_key=api_key or os.getenv("GEMINI_API_KEY"),
             http_options={"api_version": "v1alpha"},
         )
 
-        # Define the tools and system instruction
-        tools = [
-            Tool(google_search=GoogleSearch()),
-        ]
+        tools = [Tool(google_search=GoogleSearch())]
         system_instruction = Content(
             parts=[Part.from_text(text=f"{system_message}")],
             role="user"
         )
 
         config = LiveConnectConfig(
-            response_modalities=["AUDIO"],  # type: ignore
+            response_modalities=["AUDIO"],
             speech_config=SpeechConfig(
-                language_code="de-DE",
+                language_code=lang_code,
                 voice_config=VoiceConfig(
                     prebuilt_voice_config=PrebuiltVoiceConfig(
                         voice_name=voice_name,
@@ -98,7 +114,7 @@ class GeminiHandler(AsyncStreamHandler):
             tools=tools,
             system_instruction=system_instruction,
         )
-        
+
         async with client.aio.live.connect(
             model="gemini-2.0-flash-exp", config=config
         ) as session:
@@ -152,12 +168,12 @@ stream = Stream(
                 "Fenrir",
                 "Aoede",
             ],
-            value="Kore",  # Changed default to Kore
+            value="Kore",
         ),
         gr.Textbox(
             label="System Message",
             placeholder="Enter system instructions for the AI...",
-            value="Du bist ein hilfsamer Assistent, der Fragen beantwortet und bei verschiedenen Aufgaben hilft. Du kannst bei Bedarf auch im Internet suchen, um aktuelle Informationen zu finden.",
+            value="You are a helpful assistant who answers questions and helps with tasks.",
             lines=3,
         ),
     ],
@@ -172,13 +188,17 @@ class InputData(BaseModel):
 
 
 app = FastAPI()
-
 stream.mount(app)
 
 
 @app.post("/input_hook")
 async def _(body: InputData):
-    stream.set_input(body.webrtc_id, body.api_key, body.voice_name, body.system_message)
+    stream.set_input(
+        body.webrtc_id,
+        body.api_key,
+        body.voice_name,
+        body.system_message,
+    )
     return {"status": "ok"}
 
 
@@ -199,5 +219,4 @@ if __name__ == "__main__":
         stream.fastphone(host="0.0.0.0", port=7860)
     else:
         import uvicorn
-
-        uvicorn.run(app, host="0.0.0.0", port=7860)
+        uvicorn.run(app, host="0.0.0.0", port=7860)