Update app.py

app.py

@@ -6,27 +6,20 @@ from gtts import gTTS
 from audiorecorder import audiorecorder
 import speech_recognition as sr
 from pydub import AudioSegment
-from speech_recognition import Microphone, Recognizer
-import pyttsx3
 
 if "history" not in st.session_state:
     st.session_state.history = []
 
 def recognize_speech(audio_data, show_messages=True):
     recognizer = sr.Recognizer()
-    mic_list = sr.Microphone.list_microphone_names()
+    audio_recording = sr.Microphone(device_index=None, sample_rate=16000, chunk_size=1024)
 
-    if not mic_list:
-        engine = pyttsx3.init()
-        engine.say("No se encontraron micrófonos. Por favor, asegúrate de que estén conectados.")
-        engine.runAndWait()
-    else:
-        print("Micrófonos encontrados:")
-        for i, mic in enumerate(mic_list):
-            print(f"{i+1}. {mic}")
+    with audio_recording as source:
+        recognizer.adjust_for_ambient_noise(source)
+        audio = recognizer.listen(source, timeout=5)
 
     try:
-        audio_text = recognizer.recognize_google(audio_data, language="es-ES")
+        audio_text = recognizer.recognize_google(audio, language="es-ES")
         if show_messages:
             st.subheader("Texto Reconocido:")
             st.write(audio_text)
@@ -41,13 +34,14 @@ def recognize_speech(audio_data, show_messages=True):
     return audio_text
 
 def format_prompt(message, history):
-    prompt_list = ["<s>"]
+    prompt = "<s>"
 
     for user_prompt, bot_response in history:
-        prompt_list.extend([f"[INST] {user_prompt} [/INST]", f" {bot_response}</s> "])
+        prompt += f"[INST] {user_prompt} [/INST]"
+        prompt += f" {bot_response}</s> "
 
-    prompt_list.append(f"[INST] {message} [/INST]")
-    return ''.join(prompt_list)
+    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")
@@ -91,7 +85,7 @@ def text_to_speech(text, speed=1.3):
 
 def main():
     st.title("Chatbot de Voz a Voz")
-    mic = sr.Microphone(device_index=-1)
+    audio_data = audiorecorder("Habla para grabar", "Deteniendo la grabación...")
 
     if not audio_data.empty():
         st.audio(audio_data.export().read(), format="audio/wav")