Enable online speech recognition

@Brink. There is a tutorial at Enable or Disable Online Speech Recognition in Windows 10 which shows how to enable and disable online speech recognition.

Given your tutorial was written December 2018 I'm going to assume it applies to 1809, which is good, since that's the version I'm using.

I tried it but it didn't work for me, so using RegFromApp I toggled the setting in the settings app and these are the registry keys that get changed;

Code:

[HKEY_CURRENT_USER\Software\Microsoft\Speech_OneCore\Settings\OnlineSpeechPrivacy]
"HasAccepted"=dword:00000001

[HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Search]
"CortanaMUID"="MUID=c6fcb2bb327a43328968dea53d2a4303"

The MUID value is added if it doesn't exist, but turning off speech recognition does not remove it.

The first registry key is deleted when turning off online speech recognition.

Would you be so kind as to verify my findings.

So that then begs the question, what does the InputPersonalization registry value do?

Brink said:

Hello Tanya,
The group policy in option 1 below still makes the same registry change in option 2 for me in Windows 10 version 2004 build 20201.

I used Process Monitor to verify.

Enable or Disable Online Speech Recognition in Windows 10

Interesting. On 1809 the Online Speech Recognition toggle remains off after using the group policy method or the registry tweak per the tutorial for me. The only thing that toggles it on for me (apart from manually doing it), is using the registry key 'HasAccepted'.

I deleted the InputPersonalization registry value and went into group policy, enabled it there, and the InputPersonalization registry value is certainly created, but it seems to make no difference to the Online Speech Recognition - the settings toggle remains off.

Tanya

Brink said:

The tutorial you mentioned is just for enabling or disabling the feature.

You're wanting this below to actually turn it on or off.

Turn On or Off Online Speech Recognition in Windows 10

Ok, now I'm confused. I've set the HasAccepted to '1' and I can talk to Cortana. I then deleted the InputPersonalization registry value, rebooted and I can still talk to Cortana. What am I missing?

So we have an 'enable or disable' online speech recognition and a 'turn online speech recognition on or off'. lol.

UPDATE: OK, I got it. Explicitly disabling the policy or setting AllowInputPersonalization to 0 disables speech recognition. A value of 1 or removing the registry value completely are functionally equivalent.