Stem Separation and it was ME who fixed spelling in CHATGPT images

ok so after getting annoyed with Chat GPT continually breaking my code, and switching to Claude ;i have managed to get stem separation down to 4.7 secs for a 24 bit 5 min .wav track, in dos

in windows its about 10 seconds, im using the demucs protocol,and cuda enabled.

the whole CHAT GPT thing really upset me, does it actively try to annoy you? Claude was the way forward, i m total noob to all of this, but it does seem all the complicated tasks have already been written in python by someone else, and what really fucks me off is that;

I WROTE A character recognition python program thats checks and helps CHAT GPT make words sensical, so you can thank me for the fact all your pictures have the correct spelling, as it used to just produce stupid words,and ridiculous characters, remember? lol.... thanks toby....

anyway ive got a load of shit to do, i also make music, i know you all super intelligent , so how come you didnt write the OPCR program???

well love to the community...and in the words i am most happy to see, ' no problems detected in the workspace ' see you soon
ps hers the code for the OCR

import os
import pytesseract
from PIL import Image

Optional: Set path to tesseract executable if not in PATH

pytesseract.pytesseract.tesseract_cmd = r'C:\Program Files\Tesseract-OCR\tesseract.exe'

Folder containing AI-generated images

IMAGE_FOLDER = 'generated_images'

Output folder for images with "readable" text

OUTPUT_FOLDER = 'sensible_text_images'
os.makedirs(OUTPUT_FOLDER, exist_ok=True)

Function to determine if text is sensical

def is_sensical(text):
words = text.split()
# Heuristic: At least 3 "real" English words
sensible_words = [word for word in words if word.isalpha() and len(word) > 2]
return len(sensible_words) >= 3

Process each image

for filename in os.listdir(IMAGE_FOLDER):
if filename.lower().endswith(('.png', '.jpg', '.jpeg', '.webp')):
img_path = os.path.join(IMAGE_FOLDER, filename)
image = Image.open(img_path)
text = pytesseract.image_to_string(image)

    if is_sensical(text):
        print(f"✅ {filename}: '{text.strip()}'")
        # Save a copy of the image to the output folder
        image.save(os.path.join(OUTPUT_FOLDER, filename))
    else:
        print(f"❌ {filename}: Unreadable or nonsense")