Update app.py

app.py

@@ -86,33 +86,34 @@ def text_to_speech(text):
 
 def detect_voice_activity(audio_data):
     audio = pydub.AudioSegment.from_file(io.BytesIO(audio_data))
-    audio = audio.set_channels(1)  # Convertir a mono
+    audio = audio.set_channels(1) 
     signal_energy = np.sum(np.abs(audio.get_array_of_samples()))
-    threshold_energy = 5000  # Ajustar según sea necesario
+    threshold_energy = 5000 
     is_voice_active = signal_energy > threshold_energy
 
     return is_voice_active
 
 def def_main():
     st.write("Di la palabra XAMAN para empezar o DETENTE para procesar")
+
     recording = st_mic_recorder(recording_container=st.empty(), auto_recording=True)
+    audio_data = None
     
-    if recording:
-        st.write("Listening...")
-        audio_data = base64.b64decode(recording.split(",")[1])
-        audio_file = io.BytesIO(audio_data)
-        st.audio(audio_file, format="audio/wav")
-
-        is_voice_active = detect_voice_activity(audio_data)
-
-        if is_voice_active:
-            audio_text = recognize_speech(audio_file)
-            if audio_text:
-                output, audio_file = generate(audio_text, history=st.session_state.history)
-                if audio_file is not None:
-                    st.markdown(
-                        f"""<audio autoplay="autoplay" controls="controls" src="data:audio/mp3;base64,{base64.b64encode(audio_file.read()).decode()}" type="audio/mp3" id="audio_player"></audio>""",
-                        unsafe_allow_html=True)
-
-if __name__ == "__main__":
-    def_main()
+    while audio_data is None:
+        if recording:
+            st.write("Listening...")
+            audio_data = base64.b64decode(recording.split(",")[1])
+            audio_file = io.BytesIO(audio_data)
+            st.audio(audio_file, format="audio/wav")
+            is_voice_active = detect_voice_activity(audio_data)
+
+            if is_voice_active:
+                audio_text = recognize_speech(audio_file)
+                if audio_text:
+                    output, audio_file = generate(audio_text, history=st.session_state.history)
+                    if audio_file is not None:
+                        st.markdown(
+                            f"""<audio autoplay="autoplay" controls="controls" src="data:audio/mp3;base64,{base64.b64encode(audio_file.read()).decode()}" type="audio/mp3" id="audio_player"></audio>""",
+                            unsafe_allow_html=True)
+        else:
+            st.write("Waiting for voice input...")