back to first

app/chat.py

@@ -52,7 +52,7 @@ def initialize_session_state():
     if 'tts_output' not in st.session_state:
         st.session_state['tts_output'] = ""
     if 'tts_played' not in st.session_state:
-        st.session_state['tts_played'] = True  # default True supaya tidak main saat awal
+        st.session_state['tts_played'] = True  
 
 # edge-tts fallback (cadangan)
 async def generate_audio_edge(text, path, voice="id-ID-GadisNeural"):
@@ -126,20 +126,26 @@ def display_chat_history(chain):
             use_container_width=True,
         )
 
-    # Ambil input user dari teks atau suara
-    user_input = user_input_obj or stt_text
+    # Jika ada STT
+    if stt_text:
+        st.session_state.input_text = stt_text
+        st.experimental_rerun()
+
+    # Ambil input user
+    user_input = user_input_obj or st.session_state.get("input_text", "")
 
-    # Jika ada input
     if user_input:
         with st.spinner('Sedang membuat jawaban...'):
             output = conversation_chat(user_input, chain, st.session_state['history'])
 
         st.session_state['past'].append(user_input)
         st.session_state['generated'].append(output)
+        st.session_state.input_text = ""
 
-        # Mainkan TTS jika diaktifkan.
-        if st.session_state['should_speak']:
-            st.markdown(text_to_speech(output), unsafe_allow_html=True)
+        # Reset flag supaya TTS siap memutar lagi
+        if st.session_state['should_speak'] and output:
+            st.session_state['tts_output'] = output
+            st.session_state['tts_played'] = False
 
     # Tampilkan Riwayat Chat
     if st.session_state['generated']:
@@ -147,3 +153,8 @@ def display_chat_history(chain):
             for i in range(len(st.session_state['generated'])):
                 message(st.session_state["past"][i], is_user=True, key=str(i) + '_user', avatar_style="no-avatar")
                 message(st.session_state["generated"][i], key=str(i), avatar_style="no-avatar")
+
+    # Pemutaran TTS
+    if st.session_state.get('tts_output') and not st.session_state.get('tts_played'):
+        st.markdown(text_to_speech(st.session_state['tts_output']), unsafe_allow_html=True)
+        st.session_state['tts_played'] = True