Update app.py

app.py

@@ -30,7 +30,7 @@ def create_unique_wav_file(pcm_data, channels=1, rate=24000, sample_width=2):
         print(f"Error saving wave file: {e}")
         raise gr.Error(f"Could not save audio file. Error: {e}")
 
-# --- Core API Logic (Corrected Pattern) ---
+# --- Core API Logic (Corrected API Call Structure) ---
 def synthesize_speech(text, voice):
     """
     Synthesizes speech from text using the Gemini API's native TTS capabilities.
@@ -50,9 +50,8 @@ def synthesize_speech(text, voice):
         # 3. Instantiate the correct model
         model = genai.GenerativeModel(model_name="gemini-2.5-flash-preview-tts")
 
-        # 4. Construct the GenerationConfig
-        tts_config = GenerationConfig(
-            response_modalities=["AUDIO"],
+        # 4. Construct the GenerationConfig with ONLY the speech_config
+        tts_generation_config = GenerationConfig(
             speech_config={
                 "voice_config": {
                     "prebuilt_voice_config": {
@@ -62,11 +61,12 @@ def synthesize_speech(text, voice):
             }
         )
         
-        # 5. Generate content with the model and config
+        # 5. Generate content, passing response_modalities directly
         prompt = f"Say cheerfully: {text}"
         response = model.generate_content(
            contents=prompt,
-           generation_config=tts_config
+           generation_config=tts_generation_config,
+           response_modalities=["AUDIO"]  # CORRECTED: This is a direct argument
         )
         
         # 6. Extract audio data from the response structure