Update app.py

app.py

@@ -1,3 +1,4 @@
+import pyaudio
 import streamlit as st
 import base64
 import io
@@ -15,8 +16,9 @@ def recognize_speech(audio_data, show_messages=True):
     audio_recording = sr.Microphone(device_index=4, sample_rate=16000, chunk_size=1024)
 
     with audio_recording as source:
-        recognizer.adjust_for_ambient_noise(source)
-        audio = recognizer.listen(source, timeout=5)
+        list_available_microphones()
+        recognizer.adjust_for_ambient_noise(source, gain=None)
+        audio = recognizer.listen(source, timeout=5, gain=None)
 
     try:
         audio_text = recognizer.recognize_google(audio, language="es-ES")
@@ -37,10 +39,9 @@ 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"[/s] {user_prompt} [/s] {bot_response}</s> "
 
-    prompt += f"[INST] {message} [/INST]"
+    prompt += f"[/s] {message} [/s]"
     return prompt
 
 def generate(audio_text, history, temperature=None, max_new_tokens=512, top_p=0.95, repetition_penalty=1.0):
@@ -83,6 +84,15 @@ def text_to_speech(text, speed=1.3):
     modified_audio_fp.seek(0)
     return modified_audio_fp
 
+def list_available_microphones():
+    p = pyaudio.PyAudio()
+    info = p.get_host_api_info_by_index(0)
+    numdevices = info.get('deviceCount')
+    for i in range(0, numdevices):
+        if (p.get_device_info_by_index(i).get('maxInputChannels')) > 0:
+            print("Input Device id {}: {}".format(i, p.get_device_info_by_index(i).get('name')))
+    p.terminate()
+
 def main():
     st.title("Chatbot de Voz a Voz")
     audio_data = audiorecorder("Habla para grabar", "Deteniendo la grabación...")