Update app.py

app.py

@@ -4,29 +4,18 @@ import io
 from huggingface_hub import InferenceClient
 from gtts import gTTS
 from audiorecorder import audiorecorder
-import streamlit_webrtc as webrtc
 
-def recognize_speech(audio_data, show_messages=True):
+def record_audio(filename="audio.wav", duration=5):
     recognizer = sr.Recognizer()
-    audio_recording = sr.AudioFile(audio_data)
-
-    with audio_recording as source:
-        audio = recognizer.record(source)
-
-    try:
-        audio_text = recognizer.recognize_google(audio, language="es-ES")
-        if show_messages:
-            st.subheader("Texto Reconocido:")
-            st.write(audio_text)
-            st.success("Reconocimiento de voz completado.")
-    except sr.UnknownValueError:
-        st.warning("No se pudo reconocer el audio. ¿Intentaste grabar algo?")
-        audio_text = ""
-    except sr.RequestError:
-        st.error("No he recibido ningun audio. Por favor, inténtalo de nuevo.")
-        audio_text = ""
-
-    return audio_text
+
+    with sr.Microphone() as source:
+        st.subheader("Habla para grabar...")
+        audio_data = recognizer.listen(source, timeout=duration)
+
+    st.subheader("Deteniendo la grabación...")
+
+    with open(filename, "wb") as f:
+        f.write(audio_data.get_wav_data())
 
 def format_prompt(message, history):
     prompt = "<s>"
@@ -81,14 +70,16 @@ def detect_vocal_activity(audio_data):
 
     return actividad_vocal
 
-
 def main():
     if "history" not in st.session_state:
         st.session_state.history = []
 
-    if not audio_data.empty():
-        st.audio(audio_data.export().read(), format="audio/wav")
-        audio_data.export("audio.wav", format="wav")
+    start_stop_button = st.button("Iniciar/Detener Detección")
+
+    if start_stop_button:
+        record_audio("audio.wav")
+
+        st.audio("audio.wav", format="audio/wav")
         audio_text = recognize_speech("audio.wav")
 
         if not st.session_state.history:
@@ -97,23 +88,19 @@ def main():
             st.session_state.history.append((pre_prompt, output))
 
         if audio_text:
-            actividad_vocal = detect_vocal_activity("audio.wav")
-
-            if actividad_vocal.any():
-                output, audio_file = generate(audio_text, history=st.session_state.history)
-
-                if audio_text:
-                    st.session_state.history.append((audio_text, output))
-
-                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.warning("No se detectó actividad vocal.")
+            detect_vocal_activity("audio.wav")
+            output, audio_file = generate(audio_text, history=st.session_state.history)
+
+            if audio_text:
+                st.session_state.history.append((audio_text, output))
+
+            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__":
     main()