Speech and text recognition

Speech recognition from the microphone and text from the screen

Available only in the Local version of OctoWatch

  • Search by keywords or regular expressions.
  • Text will be extracted from the screenshot regardless of the application or website.
  • Speech recorded by the computer/laptop microphone will be converted to text.
  • You can create rules to perform automated actions when specific text appears on the user's screen or is spoken in the office.
Try for free for 14 days

Optical Character Recognition (OCR) on the computer screen

OctoWatch has a built-in text recognition engine that allows for real-time text detection on the screen. The OCR mechanism continuously recognizes the image on the user's computer screen and converts it into text, which is then indexed. With OCR, you can search for text even within images and videos. There are two methods of recognition available:

  • Built-in text recognition engine based on Tesseract
  • External recognition engine from ABBYY

Speech Recognition (AR)

In one of the recent updates, we also implemented speech recognition capabilities. Now, audio recorded by the built-in laptop microphone can not only be listened to but also converted into text. There are two methods of recognition available:

  • Wit.AI
  • Yandex.Cloud

Features:

  • This functionality is currently available only for users of the Local version of OctoWatch.
  • Thanks to a modified Tesseract engine, the quality of screenshot recognition has been significantly improved. No additional services or programs are needed, and there are no limits on the number of recognized images.
  • If necessary, you can connect the ABBYY cloud text recognition server.
  • The Wit.AI speech recognition engine from Facebook is currently a leader among free speech recognition solutions.
  • If needed, you can connect the Yandex engine.
  • The Recognition Server is offered as a separate free module and can be installed on a machine separate from the main OctoWatch Server.
  • You can specify the time during which the Recognition Server will operate and the maximum number of threads.
  • Recognition can be configured for some users or departments.
  • The text received as a result of recognition can be worked with just like other monitoring objects: perform keyword searches and assign automated rules-actions.
WhatsApp Logo