Text recognition
What does text recognition do?
By default the MysterSmith app is using the handwriting recognition service by Google Vision. You can send handwritings from the Doodle screen, from the WEB-doodle or from supported impression boards, and you will receive the recognized text. This text can be displayed on your PeekSmith device, supported smartwatches and all other displays (you'll need the External Displays add-on for this to work).
A better alternative to Google Vision: A.I. recognition
You can also use A.I. based text recognition, which is using either GPT by OpenAI, or Gemini by Google. To use these A.I. based recognition engines, you don't need to buy any text recognition credits from me, but you'll have to create your own OpenAI or Gemini API key.
For GPT based recognition you will have to top up your OpenAI account with at least 10 USD (paid directly to OpenAI), while using Gemini is FREE (up to a certain limit). For more information please visit this page:
mystersmith.info/ai-recognition
The prerequisites of using text recognition
To use the text recognition feature, you must have your Username and PIN set up first. If you haven't set your Username and PIN yet, go to USER SETTINGS and tap the empty Username field. There you can add any username (that is not already taken) and set any 4-6 digit number as your PIN.
You will also have to turn ON the " Enable text recognition " option in the TEXT RECOGNITION section.
Recognition credits (for Google Vision engine only)
If you use Google as the recognition engine, the recognition of one page (no matter how many text it contains) uses 1 recognition credit. Every user gets 250 credits for free when they create their user account. If you're running out of credits, you can buy recognition credits from me. You can check your credit balance in the TEXT RECOGNITION section ("Recognition credits left").
Click here to buy text recognition credits
Using A.I. text recognition (OpenAI or Gemini)
If you already have the " API + A.I. access " add-on, you can enter your own OpenAI API key or Gemini API key to be able to use A.I. based features, like text recognition, image analysis and more.
IMPORTANT: If you go with this option, you will not use Google Vision as the recognition engine, and you won't need to buy any recognition credits from me.
For details and instructions please visit this page: A.I. text recognition
Triggering text recognition
The recognition should be triggered, either automatically or by invoking an action.
To use the automatic recognition, go to the TEXT RECOGNITION section and set "Automatic recognition" to 3 seconds or more (depnding on your routine). After writing something, recognition wil trigger when you don't write anything for X seconds.
There are also some text recognition actions which can be assigned to triggers like PeekSmith buttons, Atom remote buttons, Apple Watch screen taps or swipes, WearOS watch screen taps, volume button presses etc. Some of these triggers can be found in the ACTION CONTROL section of the settings page, while some triggers can be found somewhere else (like on a device config page).
Text recognition settings explained
Recognition engine: select the preferred recognition engine (Google Vision, GPT by OpenAI, Gemini by Google, or even Offline mode)
Recognition method: select the option you need for your routine. You can recognize text/numbers, playing cards, ESP cards, timestamps, drawings etc.
Display progress message: enable this to receive a notification when text recognition is in progress
Transform Chinese/Hebrew text: enable this fi you use Chinese/Hebrew text recognition, and you want to convert the recognized text using the letters of the English alphabet
Correct spelling/names: if you enable this, the A.I. will try to correct your misspellings any typos automatically
Short card names: recognized playing cards will be converted to a format like AH, 5D, 10C, KS etc.
Automatic recognition: you can set a timer that will trigger recognition automatically X seconds after the user stopped writing on the Doodle screen or impression board. For example if you set this to 5 seconds, and the user writes a word, text recognition will automatically happen 5 seconds after they stop writing.
Show recognized text (pop-up): enable this to see the recognized text either in a pop-up message, or as a large text on the BLACK SCREEN
Clear canvas for next drawing: if you enable this, then you can perform multiple text recognitions using the same impression board (or Doodle screen), without the need to manually clear the previous writing. If you write a new word after a successful text recognition, the doodle screen will be cleared automatically as soon as you start writing
Action after recognition: if you want something to happen automatically after a successful text recognition, set this option to 1 or more seconds
Trigger action: select what should happen automatically after a text recognition
Rubik's cube colors: enable this option if you want to recognize colors on a drawing of a Rubik's cube face
CubeSmith integration: enable this option if you want to automatically awitch to CubeSmith after recognizing the colors on a cube face drawing
Numbers only: turn this option ON if you only want to recognize numbers (like for a LOTTO routine or Pi Revelations)
Pi Revelations (4 digits): if you enable this option, the app will show you the Pi Revelations numbers after recognizing a 4 digit number
LOTTO routine: enable this if you wnat to recognize LOTTO numbers
Numbers in ascending order: if you enable this, the LOTTO numbers will be arranged in ascending order
Last number as JOKER: enable this if the last number written should be a JOKER / EXTRA number (won't be included in the ascending order)
1-digit numbers with zeros: turn 1 digit numbers to 01, 05, 07 etc. format
Remove commas between numbers: return the recognized LOTTO numbers separated by SPACES instead of commas