Hello.
microsoft produces new neural sounds and offers these sounds both with azure and microsoft edge.
These new neural sounds sound really natural.
I want these new natural sounds to be used with screen readers and also to use these sounds with android.
The sounds we use in windows are very artificial.
I want the new natural sounds used with azure to be used for the visually impaired.
https://azure.microsoft.com/en-us/services/cognitive-services/text-to-speech/

Narrator has an hotkey to 'Read previous line'. It should also have a hotkey for 'Read the selected text'.

In Microsoft Calendar app narrator or any other screen reader will NOT read days of the week that newly appear on screen. For example, on the screen you have the full month of November (nothing else). You are sitting on the first week at November 6th. When you move up with the up arrow the previous week appears, but it does NOT tell you that you are now on October 30th. The only way to know is if you use the left or right arrows to move to the previous or next day. I.e 29th or 31st. Please fix!

Narrator should check the system output device when ever it is activated. If the sound device is muted, Narrator should automatically unmute it and set the volume to a value such as 50% so that it is audible. If the sound device is working properly, Narrator should do nothing and start normally. This would be a great help when trying to work with a system where the sound was muted. By pressing Control+Windows+Enter, the screen reader starts and the sound device is unmuted and set to a relatively high volume.

It would be nice to have the photos app to work or have the option to work just like the old windows movie maker since it was so blind friendly. We could make videos without asking for sighted help back then. Now we have to ask someone to make our videos YouTube ready which doesn't always work out for us.

When installing windows from a flash dive it would be nice to have narrator speak during the windows setup process. This would benefit the blind and visually impaired. Also it would be nice to be able to use narrator while in the bios.

While Narrator appears to be using some kind of online OCR engine with the image descriptions feature, it isn't very accurate and doesn't work when offline. Narrator should be able to use the built-in OCR engine in Windows 10. NVDA is using this engine and it's fantastic! I can do things like read and select inaccessible controls if they contain text or move focus to an image-based PDF and receive the text in less than 5 seconds.

Narrator and Cortana talk on top of each other when I use both at the same time.

The narrator doesn't support customizing pronunciation feature for specific words. For instance, if narrator pronounces some words incorrectly, Narrator can change the pronunciation in dictionary setting like JAWS and NVDA.

Please add an option to reset all Narrator settings to the default values. This could be helpful in many different situations, especially when troubleshooting problems or if someone wants to undo a bunch of changes and start from scratch.

If you could have more natural sounding voices for ease of access options. I am a college student, and I have difficulties with learning if I cant hear it and read it and see it, I struggle. When the voices sound synthesized and computerized, I can't focus at all i can only focus on the voice and how irritating it sounds not to be rude at all. But it is almost like nails on a chalkboard and affects me worse than not having the read-aloud option.

Some details or information on the Settings screen cannot be read by a screenreader. Some way, it has to be expressed that the data has to be accessed. It just can't be read by either Narrator, the screenreader or the JAWS mode. Was trying to find my "Last Backup date" and it only gave me buttons to backup and such, though the information was there on that same page.

Currently, when an aria live region has aria-relevant as "text", it will not read out any textual changes within the live region. You can test with this example:

<html>
<head>

<title>Aria-live test</title>

</head>
<body>

<button aria-controls="test-area" onclick="testFunction()">Click me</button>

<div id="test-area" aria-live="polite" aria-relevant="text"></div> 

<script>

  var toggle = false;


  var testFunction = function(){

    document.getElementById("test-area").innerText = toggle ? "Even newer text" : "New text";

    toggle = !toggle;

  }

</script>

</body>
</html>

NVDA reads the changes in this example out fine but narrator does not.

Only narrator can read candidate item of Chinese input method. Other screen reader cannot get candidate item of Microsoft Pinyin through ITfUIElementSink