Update app.py

app.py

@@ -1,42 +1,12 @@
 import streamlit as st
 import base64
 import io
-import pyaudio
-import wave
-import numpy as np
-from gtts import gTTS
 from huggingface_hub import InferenceClient
+from gtts import gTTS
+from audiorecorder import audiorecorder
+import speech_recognition as sr
 
-def record_audio(filename="audio.wav", duration=5):
-    st.subheader("Habla para grabar...")
-
-    p = pyaudio.PyAudio()
-    default_input_device_index = p.get_default_input_device_info()['index']
-    
-    stream = p.open(format=pyaudio.paInt16,
-                    channels=1,
-                    rate=44100,
-                    input=True,
-                    frames_per_buffer=1024)
-
-    frames = []
-
-    for i in range(0, int(44100 / 1024 * duration)):
-        data = stream.read(1024)
-        frames.append(data)
-
-    stream.stop_stream()
-    stream.close()
-    p.terminate()
-
-    wf = wave.open(filename, 'wb')
-    wf.setnchannels(1)
-    wf.setsampwidth(pyaudio.PyAudio().get_sample_size(pyaudio.paInt16))
-    wf.setframerate(44100)
-    wf.writeframes(b''.join(frames))
-    wf.close()
-
-def recognize_speech(audio_data):
+def recognize_speech(audio_data, show_messages=True):
     recognizer = sr.Recognizer()
     audio_recording = sr.AudioFile(audio_data)
 
@@ -45,16 +15,28 @@ def recognize_speech(audio_data):
 
     try:
         audio_text = recognizer.recognize_google(audio, language="es-ES")
-        st.subheader("Texto Reconocido:")
-        st.write(audio_text)
-        st.success("Reconocimiento de voz completado.")
-        return audio_text
+        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?")
-        return ""
+        audio_text = ""
     except sr.RequestError:
         st.error("No he recibido ningun audio. Por favor, inténtalo de nuevo.")
-        return ""
+        audio_text = ""
+
+    return audio_text
+
+def format_prompt(message, history):
+    prompt = "<s>"
+
+    for user_prompt, bot_response in history:
+        prompt += f"[INST] {user_prompt} [/INST]"
+        prompt += f" {bot_response}</s> "
+
+    prompt += f"[INST] {message} [/INST]"
+    return prompt
 
 def generate(audio_text, history, temperature=None, max_new_tokens=512, top_p=0.95, repetition_penalty=1.0):
     client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
@@ -73,7 +55,7 @@ def generate(audio_text, history, temperature=None, max_new_tokens=512, top_p=0.
         seed=42,
     )
 
-    formatted_prompt = f"<s>[INST] {audio_text} [/INST]</s>"
+    formatted_prompt = format_prompt(audio_text, history)
     stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=True)
     response = ""
 
@@ -95,24 +77,23 @@ def main():
     if "history" not in st.session_state:
         st.session_state.history = []
 
-    start_stop_button = st.button("Iniciar/Detener Detección")
+    audio_data = audiorecorder("Habla para grabar", "Deteniendo la grabación...")
 
-    if start_stop_button:
-        record_audio("audio.wav")
-
-        st.audio("audio.wav", format="audio/wav")
+    if not audio_data.empty():
+        st.audio(audio_data.export().read(), format="audio/wav")
+        audio_data.export("audio.wav", format="wav")
         audio_text = recognize_speech("audio.wav")
 
         if not st.session_state.history:
             pre_prompt = "Te Llamarás Chaman 4.0 y tus respuestas serán sumamente breves."
             output, _ = generate(pre_prompt, history=st.session_state.history)
             st.session_state.history.append((pre_prompt, output))
-
+        
         if audio_text:
-            output, audio_file = generate(audio_text, history=st.session_state.history)
+            output, audio_file = generate(audio_text, history=st.session_state.history)  
 
-            if audio_text:
-                st.session_state.history.append((audio_text, output))
+            if audio_text:  
+                st.session_state.history.append((audio_text, output))  
 
             if audio_file is not None:
                 st.markdown(