Update chat_app_remote.py

chat_app_remote.py

@@ -16,18 +16,6 @@ custom_css = """
 .gradio-container {
     justify-content: flex-start !important;
 }
-.send-btn {
-    width: 140px;
-    height: 40px;
-    min-width: 140px;
-    padding: 0;
-}
-.centered-col {
-    display: flex;
-    flex-direction: column;
-    align-items: center;
-    justify-content: center;
-}
 """
 
 def create_frontend_demo():
@@ -64,14 +52,13 @@ def create_frontend_demo():
                     placeholder="Start chatting with Aira..."
                 )
                 
-                # Use a Column with a custom class to center its items
-                with gr.Column(elem_classes="centered-col"):
+                # Place just the text box (removing the send button)
+                with gr.Column():
                     msg = gr.Textbox(
                         show_label=False,
                         placeholder="Enter text and press enter",
                         container=True
                     )
-                    send_btn = gr.Button("➤", elem_classes="send-btn")
                 
                 audio_output = gr.Audio(
                     label="Aira's Response",
@@ -87,7 +74,7 @@ def create_frontend_demo():
                         label="Audio Input",
                         streaming=False
                     )
-
+                    
             with gr.Tab("Options"):
                 with gr.Column():
                     session_input = gr.Textbox(
@@ -104,7 +91,7 @@ def create_frontend_demo():
                     You can talk to her in English or Japanese, but she will only respond in Japanese (Subs over dubs, bros) ask her to give you a Subtitle if you can't talk in Japanese. <br>
                     
                     The majority of the latency depends on the HF's inference api.
-                    LLM is not fine-tuned or optimized at all. the current state of conversational off-the-shelf japanese LLM seem to be less than remarkable, please beware of that.
+                    LLM is not fine-tuned or optimized at all. the current state of conversational off-the-shelf japanese LLM seems to be less than remarkable, please beware of that.
                     
                     1. Enter your Session ID above or leave blank for a new one
                     2. Click 'Set Session ID' to confirm
@@ -123,12 +110,6 @@ def create_frontend_demo():
             inputs=[msg, chatbot, session_id_state],
             outputs=[msg, chatbot, audio_output, session_id_state, session_display]
         )
-        # Also allow clicking the send button
-        send_btn.click(
-            respond,
-            inputs=[msg, chatbot, session_id_state],
-            outputs=[msg, chatbot, audio_output, session_id_state, session_display]
-        )
 
         def set_session(user_id):
             result = client.predict(
@@ -150,30 +131,22 @@ def create_frontend_demo():
                             
             try:
                 sample_rate, audio_array = audio_data
-                
                 with tempfile.NamedTemporaryFile(suffix='.wav', delete=True) as temp:
                     wavfile.write(temp.name, sample_rate, audio_array)
-                    
                     audio = {"path": temp.name, "meta": {"_type": "gradio.FileData"}}
-
-                    # Get the result while the temporary file still exists
                     result = client.predict(
                         audio,
                         history,
                         session_id,
                         api_name="/handle_audio"
                     )
-                    
-                    # Unpack and construct the display text
                     audio_path, new_history, new_session_id = result
                     display_text = f"Current Session ID: {new_session_id}"
-                
                     return audio_path, new_history, new_session_id, display_text
-
             except Exception as e:
                 print(f"Error processing audio: {str(e)}")
                 import traceback
-                traceback.print_exc()  # This will print the full error traceback
+                traceback.print_exc()
                 return None, history, session_id, f"Error processing audio. Session ID: {session_id}"
             
         audio_input.stop_recording(