Small Python script to animate character depending on voice volume (used with OBS virtual cam in school calls)

vtuber.py

from PIL import Image
import aubio
import pygame
import pyaudio
import numpy as np
import glob

PATH                 = "happyboy1"
METHOD               = "default"
PERIOD_SIZE_IN_FRAME = 2048 // 2

pA = pyaudio.PyAudio()
mic = pA.open(
    format = pyaudio.paFloat32, channels = 1,
    rate = 44100, input = True,
    frames_per_buffer = PERIOD_SIZE_IN_FRAME
)

img_size = 0, 0

with Image.open(glob.glob(f"{PATH}/*.png")[0]) as img:
    img_size = img.size

pygame.init()

size = width, height = img_size

clock = pygame.time.Clock()
screen = pygame.display.set_mode(size)

avatars = []

for image in glob.glob(f"{PATH}/*.png"):
    avatars.append(pygame.image.load(image))

step_prev = 0
step = 0
steps = len(avatars) - 1

running = True

screen.fill((255, 255, 255))
screen.blit(avatars[0], (0, 0))
pygame.display.update()

while running:
    for event in pygame.event.get():
        if event.type == pygame.QUIT:
            running = False
            break

    # READ MIC
    data = mic.read(PERIOD_SIZE_IN_FRAME)
    samples = np.frombuffer(data, dtype = aubio.float_type)
    volume = (np.sum(samples * samples) / len(samples)) * 1000

    if (volume > 25):
        step = 4
    elif (volume > 19):
        step = 3
    elif (volume > 12):
        step = 2
    elif (volume > 6):
        step = 1
    elif (volume <= 6):
        step = 0

    #print(volume)

    # Only draw if something needs to be updated
    if (step_prev != step):
        step_prev = step
        screen.blit(avatars[step], (0, 0))
        pygame.display.update((260, 370, 470, 210))

    clock.tick(50)

pygame.quit()
quit()