SeeingAI

Introduction

SeeingAI is a free speaking camera app for blind and low vision people which is developed by a team at Microsoft. Some screen readers don't speak its name very well; it should be spoken as Seeing A I. A I stands for Artificial Intelligence, the technology that lies at the heart of this app.

A note for low vision users

This web page is written for Voiceover users but should be easy to adapt for non-VoiceOver use. You select a channel by tapping its icon at the bottom of the screen and you just single tap other buttons where a VoiceOver user double taps. VoiceOver users have to swipe to move through items on the screen to discover what's there. You can just tap the item you want. The icons at the bottom of the SeeingAI screen may be difficult for low vision uses to interpret but their function will be spoken when they are tapped.

SeeingAI features

SeeingAI initially has 3 tabs at the bottom of the screen that can be navigated by horizontal swipes. They are Reed, Describe and More. Navigate to the tab you want to use and then double tap to select it. (please note that I have deliberately mis-spelt the word R E A D to encourage screen readers to speak it corretly and not speak it as red.)

The Reed tab offers powerful features for reeding both printed and handwritten text.

The Describe tab offers AI descriptions of what the camera sees, face recognition and a feature to have your Photos described.

The More tab reveals additional features which VoiceOver users will hear as "channels". VoiceOver users can swipe up or down to navigate the list of available channels. You shouldn't double tap once you navigate to the channel you want. It is already selected.

You'll find detailed advice on using SeeingAI's most popular features followng the installation instructions below.

Installing SeeingAI

SeeingAI should be installed from the App Store. The first time you open the app it will ask for permission to access the camera and maybe for other permissions, which should all be granted.  It will also provide introductory information about the app. On this web page, I'll assume that you have listened to all the introductory information and that the app is fully ready for use and that you are using VoiceOver, although the app will function fully and speak its output for non VoiceOver users.

Using SeeingAI

You can start the app by asking Siri to "open Seeing A I" or by locating and double tapping the app icon.

The app will probably start up using the last feature you used or the Reed feature if this is your first use. To explore the features available, locate the tab bar near the bottom of the screen and swipe left or right to locate the tab you want to use and double tap to select it.

While you are selecting a tab your iPhone or iPad may spot text or objects and start speaking. If this is a problem, place your hand in front of the camera which is on the back of your phone, near the top right corner. Be sure to avoid touching the camera lens since fingerprints can significantly impair performance.

Most popular SeeingAI features in more detail

Reed

Select the Reed tab and then point the camera at printed or handwritten text and it will be spoken to you. This can be text on a sheet of paper, a signpost, a package, whatever. It doesn't matter if the text is upside down or on its side. A letter may have just come through the door and you want to check if it is for you or your partner. Just point the camera at the letter to hear the address . Don't hold the phone too close to the printed text. You will get a feel for the best distance to hold things from the camera with a little practice. Try to hold the phone steady. If the phone's view of the text changes significantly, it may start reeding from the top again. I find that it is often best to point the phone down at the floor when speech has started to avoid unwanted restarts.

Reed generally does a good job of reading small amounts of text but sometimes has trouble dealing with text in multiple columns or multiple blocks. UK £ signs are sometimes recognised incorrectly. Don't be surprised if you hear prices in dollars, euros or perhaps prices prefixed by the letters E or L.  The good news is that this channel works entirely locally on your phone, so will work even when you have no wi-fi or mobile data signal, as is often the case in large buildings.

The "Take Picture" button

If you have a printed or hand written document to read, or perhaps text on a museum wall, you wil probably get better results using the circular "take Picture" button above the middle of the tab bar. This takes a picture of your document and sends it via the Internet for processing.  This feature won't work if you are not on wi-fi or you are out of range of mobile data. If you can see well enough to hold the phone so that the text is fully visible on the screen, position the phone and then locate and double tap the take picture button. If you can't position the phone  using your vision help is at hand. 

The Document Alignment button

There is a "Document Alignment" button to the left of the large circular take picture button. Oddly, a VoiceOver user will have to swipe right, not left, from the take picture button to locate the Document Alignment button. Double tap the Document Alignment button to turn on this feature and you'll get help to position the phone and the picture will be taken automatically when the position is correct. Place the document, if possible, on a contrasting surface such as a dark table top. Rest the phone on the document with its  camera near the centre of the document and roughly aligned with the edges of the page. Now hold the phone level and raise it slowly. SeeingAI will tel you which edges of the document it sees and once you've positioned the phone well enough for it to proceed, it will instruct you to "hold steady". You'll hear a shutter sound and then a little jingle. Once this is started, you can lower the phone and wait until the jingle stops playing. The text of the document will now be on your phone's screen, waiting for you to read it with the usual VoiceOver gestures.  VoiceOver's focus will be on the back button at top left of the screen. You can swipe right or slide your finger down the screen until you get to a part of the document that you want to hear.  Alternatively you could turn the rotor to headings and swipe down to skip between headings detected by Seeing AI. Start continuous reeding with the usual two finger swipe down and pause and restart reeding with a two finger tap. When you have finished, feel near top left of the screen for the "back button" and double tap. You are then ready to read another document or to select another SeeingAI feature.

As an alternative to using VoiceOver gestures to hear the document spoken you will find a  play button near bottom left of the screen. This will start continuous reeding of the document and will then become a stop button. If you use the play button a strip of buttons will appear above it. These are skip back, pause or resume, skip forward and stop.  Skip forward and skip back can be helpful since the document channel usually identifies headings in documents and will skip to the next or previous heading. It's also worth noting that sometimes, when a document is spoken using the play button, you may hear more helpful information about the document's layout than when having the document spoken by VoiceOver commands.

If you want to scan multiple pages into a single document you will find an "Add Page" button near the top right of the screen after the first page is scanned. I often use this in concerts and at the theatre to scan the entire programme before I hand it back to my wife to read!

Near bottom right of the screen you will find an "ask SeeingAI" button. Double tap this and you will be editing a text field where you can type or dictate a question about the document. As an example if you just scanned a menu, rather than listening to the whole menu you might want to ask if there are vegan options. Send your question using the send button at the right above the keyboard. The responses can be extremely good but be aware that they are generated by artificial intelligence and may contain errors. Use the answers to your questions with caution.

It is possible to save both the photo of the document and the text that Seeing A I recognised using the share button at bottom right of the screen which displays the text of the document. The text may be saved as plain text or as html, which preserves the detected layout and headings.

In early 2025 this feature seems still to be confused by some complex text layouts and doesn't always handle newspaper columns or documents with text bubbles very well. If you get poor results you may get better results on complex layouts using the Scene channel in SeeingAI or the Be My AI feature in the Be My Eyes app which both use a fuller AI approach to scanning the document. You could take advantage of SeeingAI's assistance in positioning the document for its scan by using the Share button to share the photo you have taken and then select Describe with Be My Eyes from the list of share options. This will probably give you a summary of the document but it will probably respond to a request like "read the full document" or "read the main part of the letter".

The more button, visually 3 dots in a circle, at top right of the page offers several features including a "page info" button which will tell you the orientation of the scanned page, whether it is upside down or on its side. This can be helpful if you want to file the printed page for others or staple several pages together.

Describe

When you select the Describe tab the screen will have a circular Take Picture button above the middle of the tab bar. To its left is a Face Recognition button and to its right is a Switch to Front Camera button and above that is a Browse Photos button.  These can all be discovered by VoiceOver users with left and right swipes or by exploring the screen. The Take Picture button can be used to get a detailed AI description of whatever the camera is pointing at and you can use the Switch to Front Camera button if you want to know something about yourself, perhaps about your clothing.  When you use the Take Picture button you'll hear the jingle and soon after you will hear the first part of a description of the scene. You may need to swipe right to persuade VoiceOver to complete the description. If you want to ask questions about the scene, use the "Ask SeeingAI" button at bottom left of the screen. Finally, swipe left to find the back button and double tap with one finger to return to the main Describe screen. 

Do remember that the scene descriptions are generated by artificial intelligence. The descriptions are often very helpful but be aware that there may be mistakes and you should use the descriptions with caution.

Browse Photos

The Browse Photos button on the main Describe screen is used to get SeeingAI to provide AI descriptions of your Photos. This feature can also be found using the menu button which is at top left of most SeeingAI screens. Browse Photos displays thumbnails of your photos and will initially announce the date the photo was taken and you may hear a brief description of the photo. If you double tap,a detailed AI description of the photo is generated. You will need to use VOiceOver gestures to hear this spoken. There will also be a "Ask SeeingAI" button at bottom left of the screen to allow you to ask questions regarding the photo.

Browse Photos can also describe videos. To generate a description, double tap a video and then use the Describe button near bottom left of the screen. Once it is generated, the description will be inserted as the video plays. 

More

More, the third tab on the main SeeingAI screen will display an additional set of fetures  in a "channel" bar below the main tab bar. These may be navigated by touching the channel bar and then swiping up or down with one finger.  The channels on offer will depend on your iPhone or iPad model and also on Seettings in the SeeingAI menu.I'll leave you to discover these for yourself with the exception of the Product channel which I outline below.

product

If the Reed tab wasn't able to identify a product, then select the product channel to find and look up a barcode. If the product is in a box, you may need to search all the faces for a barcode.  Don't hold the phone too close to the product. Point the camera at the package and you'll hear a pulsing sound that speeds up when you are close to a barcode.  Once the barcode has been scanned, you'll hear a musical jingle and eventually you should hear the product description. Once you've heard the product description, swipe left to find the "close, button" and double tap to return to the product channel and, if you wish, scan another product. You will need to rotate cans slowly in front of the camera to locate barcodes.

This channel also automatically looks for a new type of code, called an enhanced QR code. These codes haven't appeared yet on many products in the UK but they do have the advantage that they can be spotted by your phone at a greater distance than barcodes or conventional QR codes. SeeingAI will also speak the distance of enhanced QR codes. They should also take you to accurate information unlike barcodes, which will sometimes not be recognised in the database of barcodes that SeeingAI uses..

Describing items in other apps

Photos, videos, inaccessible PDFs in apps such as Photos, Mail, Messages or Files can be shared with SeeingAI to have them described. Use the usual Share button and scroll down to "Recognise with SeeingAI which should take you to the appropriate feature in SeeingAI.

Siri shortcuts

It is possible to start SeeingAI in a channel of your choice with a command to Siri. In order to do this you will need to set up some Siri shortcuts which are provided by the app. You need do this only once and it's simple to do. Open SeeingAI and locate and double tap the Menu button near top left of the screen. Swipe right to Settings and double tap. Swipe right to "configure Siri shortcuts button" and double tap. Swipe right through the available options and double tap to select one you want with a double tap. That will take you to a screen informing you that the shortcut has been added with a default phrase to use for Siri and if you like that default phrase, locate the Done button and double tap. If you prefer a different voice phrase then locate and double tap the change voice phrase button, double tap and then dictate your preferred phrase. Next, locate and double tap the Done button to accept your new voice phrase and then locate and double tap the final Done button to set up the shortcut.

There are no Siri shortcuts provided to start in either the Reed or Describe tabs. If you set up Siri Shortcuts in a previous version of SeeingAI for features now grouped in the Reed and Describe tabs they may still operate but shortcuts for the former short text and document channels will probably both simply open the Reed tab without Document Alignment selected. If your old shortcuts no longer work as expected I suggest you delete them in the Shortcuts app.

Back to navigation page