Microsoft released the Recall feature for users of Windows Insider builds. I gave the feature a try and made an audio recording with a roughly edited transcript. The short version of my experience is that I can get around Recall but am still trying to figure out how to read search results effectively with a screen reader, hence the title of No Big Finish.
Since this blog post is about an AI feature, I’ll offer an unedited version of Copilot’s summarization of my transcript as well. Overall the summary is reasonable but in my opinion misstates what I said about the Click To Do feature. I said I hadn’t figured out how it is supposed to work.
Perhaps because it was at the end, the summary also missed my biggest point entirely about still trying an effective way to use the search functionality of Recall and reading the meaningful text easily. Here is the Copilot summary:
The document is a transcript of a demonstration by Kelly on the Microsoft feature called “Recall.” This feature captures screenshots of your screen and allows you to search for information within those screenshots. Kelly explains that Recall was released as a test version for Microsoft Insiders on November 22nd. The demonstration focuses on the accessibility aspects of Recall, particularly how it works with screen readers like NVDA, Narrator, and JAWS.
Kelly describes the setup process, which includes signing in with Windows Hello, and demonstrates how to navigate the Recall interface using keyboard commands and screen reader feedback. The document details various functionalities of Recall, such as saving snapshots, opening Recall from the system tray, and navigating through captured images. Kelly also mentions some issues and limitations encountered during the demonstration, such as certain screen reader commands not working as expected and the need for further development to improve accessibility.
Kelly starts by explaining the initial setup, which involves a brief opt-in experience where the user grants permission for the computer to take screenshots. This process takes about 20 to 30 seconds and requires signing in with Windows Hello. Kelly notes that each time Recall is opened or used, the user is prompted to sign in with Windows Hello.
The demonstration continues with Kelly navigating the Recall interface. Kelly shows how to open Recall from the system tray and describes the various options available, such as saving snapshots, opening Recall, pausing the feature, and accessing more settings. Kelly highlights that the timeline feature in Recall displays the captured images, referred to as “memories,” and sessions, which represent the time periods during which the computer was used.
Kelly also points out some issues with the screen reader’s ability to read the titles of the snapshots automatically. Kelly demonstrates how to navigate through the list of snapshots and mentions that sometimes the titles do not read correctly on the first attempt. Kelly suggests using the home and end keys to navigate through the list and get the titles to read properly.
Additionally, Kelly explores the commands available for each snapshot, such as opening the associated application (e.g., Outlook), copying the snapshot, deleting it, and accessing more options. Kelly notes that some features, like the “Click to do” button, are not fully functional yet and require further development.
Kelly also demonstrates how to get more details about a captured image using Narrator’s command for image descriptions. However, the output is often gibberish due to the limitations of the OCR (Optical Character Recognition) technology used. Kelly suggests using other solutions like JAWS, Google’s Gemini, or Be My Eyes for better image descriptions.
Overall, the document provides an insightful look into the Recall feature, its current capabilities, and areas for improvement from an accessibility perspective. Kelly’s demonstration highlights the potential of Recall while also acknowledging the need for further development to enhance its functionality and accessibility.
Comments