use same model instead of instantiating a new one to reserve vram

main.py

@@ -1,6 +1,7 @@
 import pyautogui as pg
 from pynput import keyboard
 import speech_recognition as sr
+from faster_whisper import WhisperModel
 from string import punctuation
 from slang import replacements
 import re
@@ -8,22 +9,26 @@ import subprocess
 from time import sleep
 from mss import mss
 import numpy as np
+from io import BytesIO
 import sounddevice # turns off alsa error logging
 
 r = sr.Recognizer()
 
+model = WhisperModel("distil-small.en", device="cuda", compute_type="float16")
+
+print("Testing Sound")
 with sr.Microphone() as source:
     r.adjust_for_ambient_noise(source, duration=3)
+print("ready!")
 
 def recognize_text() -> str:
 
     with sr.Microphone() as source:
-        print("Say something!")
         audio = r.listen(source)
 
-        result = r.recognize_faster_whisper(audio, model="distil-small.en", beam_size=5, language="en", condition_on_previous_text=False)
+        results, _ = model.transcribe(BytesIO(audio.get_wav_data()), beam_size=5, language="en", condition_on_previous_text=False)
 
-    return result
+    return " ".join([segment.text for segment in results])
 
 def chat_type():
 
@@ -49,13 +54,13 @@ def on_press(key):
     if key is not keyboard.Key.home:
         return
     
-    print("triggered!")
+    print("Listening...")
     command = recognize_text()
 
     print(f"Heard: {command}")
 
     # cleanup command
-    command = command.lower()
+    command = command.lower().strip()
     for char in punctuation:
         command = command.replace(char, '')
 
@@ -86,7 +91,7 @@ def on_press(key):
         # sleep(0.074)
         # pg.keyUp("enter")
 
-    elif any(keyword in command for keyword in ["maximum", "pulse", "balls", "remove", "eliminate", "murder", "goon"]):
+    elif any(keyword in command for keyword in ["maximum", "pulse", "balls", "remove", "eliminate", "murder", "goon", "obliterate", "delete", "piss"]):
         print("MAXIMUM PULSE!!!!")
         pg.keyDown("q")
         sleep(0.032)