Using TopOCR as a Reader for the Blind

We are attempting to create an application that allows blind users to use an ordinary smartphone as a "reading device" when it's connected to their PC. Our software is currently missing a lot of features that would make it easier and more convenient for blind users to use, but because of the high cost of competitive solutions, we feel there's some value in notifying potential users about our efforts now. The amount of feedback that we gain from people who try out our software for this purpose will be the main factor that determines how much effort we put into this project to improve it. We are actively soliciting the advice and comments of blind users. The future of this project is in your hands!

What it Does

Using our software as a "reader" involves nothing more than using the command-line inteface to process JPEG pictures stored in a smartphone. With just a simple 3 line batch file, (see below) you can process all the JPEG files your camera keeps in an image directory into a single text file, and then "speak" the text output file it created with the "Flite" speech synthesizer. Because of it's simplicity, it can be re-programmed by almost anyone to do new things, like saving the audio file, converting it to another format, etc.

Setup

  • Download and install TopOCR
  • Download the Flite speech synthesizer from here and copy it to to your TopOCR directory, generally C:\program files\topocr
  • Download and modify if necessary the readall.bat batch file and the readone.bat batch file and put them into your TopOCR directory
  • This is a good time to take a few sample document images with your smartphone and then plug your smartphone into your PC.
  • The batch files contain a default "THRESHOLD" setting for the Sony Ericsson K750i Smartphone, these are the three numbers after the "-THRESH" switch, you may need to modify these settings for your Smartphone, it'd be easier for a sighted person to change these using the GUI. When you select the "Binarize" function under the Image menu, you will get a preview of how well your -THRESH settings will work. If the image is sharp and crisp without noise or broken characters, you likely have a good setting, experiment a bit to get the best results, and then copy these three numbers to your readall.bat and readone.bat batch files.

Usage

We recommend you have at least a 2 MP camera. You should read our tutorial to understand the issues and techniques involved in capturing documents with a digital camera. It can take a lot of practice and patience to develop the ability and skill necessary to perform document capture with a digital camera. So be prepared to spend some time on this. Once you have setup everything you need to open a "command prompt" window by hitting the "windows" key on your keyboard and scrolling down and selecting the correct "command line" icon. Before you run the program you will need to determine which directory on your smartphone is used to store images. Then just type "cd c:\program files\topocr" or to wherever your TopOCR installation directory is, and type "readall pathofyourimagefiles", and then when the batch file has finished processing, you will hear the audio output on your headphones or speaker!

The Future

  • Auto-rotate, so images can be in any orientation, upside down, etc.
  • Higher quality SAPI TTS voices and voices other than English
  • Automatic installation
  • Automatic Threshold calculation and setting
  • Audio feedback such as, "sorry not enough light in this image", etc
  • "Auto-launch" of TopOCR whenever the smartphone is plugged in
  • And whatever other ideas our readers can come up with

We are currently working on a mobile version of our software which has the OCR engine running inside a web browser. You could take a picture of a menu in a French Restaurant then OCR the image and translate it from a web hosted translation server and the Text To Speech Synthesizer will pronounce everything in your own language. Then at long last you'll finally know what "glacer la crème" is!