Changed the method of getting text from using pyautogui to grab a part of the screen and the using pytesseract to try and recognize text from the image. The major downsides from this approach are that:
- The Spotify window needed to be in front of the screen, at the same place
- The user needed to manually input / choose where to take the text
- Needed tesseract-ocr to be installed
- Wouldn't work with bigger titles or artist groups
- Wouldn't recognize non-english text
- Would get the text wrong about 30% of the time
The new version/release uses pymem to fetch the song url from the Spotify process, fetching the html of the Spotify page related to that song, and parsing the html to get the title and artist.
Major improvements include:
- Doesn't require the user to set up boxes for image recognition
- Doesn't require the user to install tesseract-ocr
- Doesn't require Spotify to be at the same place
- Less cluttered GUI
- Accurate more times
- Works with bigger titles and artist groups
- Works with different languages
- Doesn't need tesseract-ocr to be installed