image-to-sound-python- VS Typewriter-OCR-TintypeText

A Python script to convert a picture to sound

For a link to my typewriter TintypeText optical character recognition (OCR) Github repository, please consult: https://github.com/LPBeaulieu/Typewriter-OCR-TintypeText

Using some sort of front end like gscan2pdf, tesseract can output one giant text file with all pages. It is also very accurate, so long as your typewriter uses a relatively standard typeface (it might have trouble with script) and your machine is relatively clean. There's also this GitHub project that intends to create a perfect typewriter OCR for every typewriter by training a new OCR model for every specific machine, but this might now work in OP's case because the typewriter may be long gone by now. It will take a while to feed the pages, and editing will be required, but it would probably be cheaper assuming OP has access to a scanner and editing will be required unless you pay someone else to edit.

Dakin’s full strength solution (0.5% aqueous bleach) was prepared by diluting one part of 5% bleach in 9 parts of tap water. This solution was added to a one ounce amber glass eye dropper vial. A thin bead of the Dakin’s solution was then added along some lines of the text shown in Figure 1, using the eye dropper pipette cap. Within a few minutes, the text was erased, and the page was placed in between two layers of blotting paper, before the bleach had dried out. Some books were then placed on the paper stack to avoid excessive wrinkling of the paper as it dried out over the next few minutes. Figure 3 depicts the segmentation image of the erased page in TintypeText (typewriter OCR application, see my github: https://github.com/LPBeaulieu/Typewriter-OCR-TintypeText). We can observe that the erased lines aren’t registered by the OCR application, thus demonstrating the effectiveness of Dakin’s solution in erasing the typewritten text.

See my github repositories and Youtube instructions videos for Make Your Own Gamebook (github: https://github.com/LPBeaulieu/Make-Your-Own-Gamebook/blob/main/README.md, Youtube: https://www.youtube.com/watch?v=-IXC6wVwy4c) and TintypeText (github: https://github.com/LPBeaulieu/Typewriter-OCR-TintypeText, Youtube: https://www.youtube.com/watch?v=FG9WUW6q3dI&list=PL8fAaOg\_mhoEZkbQuRgs8MN-QSygAjdil)

Visit my github page (https://github.com/LPBeaulieu/Typewriter-OCR-TintypeText) for detailed instructions (with many illustrations) on how to train your own model on your typewriter (no coding experience needed) or how to use the models that I have trained on mine.