Note: These features are currently only available to people using Oculus in the United States with English set as the default language.
How do I use Voice Commands in my Oculus device?
If you have an Oculus Quest or Oculus Go, you can use your voice to navigate, control your device and find new apps and games. Once you've turned on Voice Commands
there are are three ways you can activate them:
Click the Voice Commands button on the Home menu.
Double press the Oculus button on the right controller.
Use the wake word and say "Hey Facebook".
If you choose to enable the Voice Command Controller Shortcut available on Oculus Quest, you can double press the Oculus controller button to use voice commands. Enabling this shortcut will disable the passthrough shortcut if you previously enabled it. You can manage these shortcuts in the Experimental Features section of Settings.
When is Oculus listening for the wake word "Hey Facebook"?
When "Hey Facebook" is enabled, your Oculus device only sends "Hey Facebook" voice interactions to Facebook servers if your microphone is on and it detects someone saying "Hey Facebook". When Oculus hears the wake word, it will start to record your voice interaction and send it to our servers in real-time to respond to your request. You'll know when Voice Commands are activated through a visual confirmation in-VR.
IIf you choose to enable "Hey Facebook", Oculus won't listen for the wake word when your microphone is off, your device is asleep, or your device is off.
What information do you collect when I use Voice Commands?
When you use Voice Commands through any activation method (Universal Menu Button, Oculus button double press or “Hey Facebook” wake word), we record your voice and create a computer-generated transcription of the recording which is then processed on our servers. We also collect related data such as the hardware version of your device and the length of audio related to your interaction. Depending on where you use Voice Commands, the microphone may pick up other sounds in the immediate area beyond your voice -- this may include ambient noise or nearby background conversations.
To help Voice Commands better understand and fulfill requests, text transcripts and related data about your voice interactions are stored by default. We can use machine learning and trained reviewers to process stored transcripts and related data to help make Assistant smarter and more accurate.
What you can do with your transcripts:
View and delete them in Settings.
Turn off storage in Settings, and still use Voice Commands.
Users can opt-in to storing their audio recordings about your voice interactions to improve Voice Commands. When voice recording storage is enabled, audio recordings about your voice interactions are stored to help Voice Commands better understand and fulfill requests. We can use machine learning and trained reviewers to process your stored voice recordings to improve Voice Commands. You can turn this off in Settings at any time. Even with storage off, you can still use Voice Commands.
Note that we will record your voice if you unintentionally trigger the Voice Commands button. If our systems detect that you did not intend to trigger Voice Commands through any activation method, we label these recordings as "false wakes" and delete them from our servers within 90 days of detection.
How do I control how you handle my voice commands?
Product Users can access (listen to and view) and delete stored transcripts and audio recordings of voice interactions in settings. Stored voice data (which includes audio, transcripts and related data about voice interactions) can be accessed and downloaded in the Download Your Information (DYI) tool.
Additionally, you can view your voice activity in the follow places:
The Account Settings of your Oculus app
Voice Activity settings on the Oculus website
The View Your Information Tool in your Privacy Center
Only users who have not opted out of transcript and related data storage will have the option to opt in to storing audio recordings of their voice interactions (voice recordings) to improve Voice Commands functionalities.
You can still use voice commands even if you turn off storage. When storage is turned off, we won’t keep recordings or transcripts of your voice commands, which means you won't be able to view, hear or delete them. And we won’t use those voice commands to improve our voice services and they won’t be reviewed by trained reviewers. We’ll still log system activity, such as the timestamp for when a voice command was made and the general category of the voice interaction. For example, if you said, “search for a dinosaur app?,” we would simply categorize your question as “search.” This information is necessary for product functionality. Any voice command made before you turned off storage will still be stored unless you delete them.
Note that if you use voice commands to open websites on Oculus browser, we will keep a record of your commands, including the URL, unless you turn off Storage. This is true even if you opted out of data collection on Oculus Browser. Visit our Privacy Information page
for more information about Oculus browser processing.
How do you use my voice commands?
We primarily use recordings, transcripts and related data of your commands to process and carry out your request. When storage of your voice commands is turned on, we also use stored recordings and transcriptions of your commands to help make Facebook’s voice services smarter, more accurate, and better for everyone. When voice storage for product improvement is on, we use machine learning and trained reviewers to process information to improve, troubleshoot and train our speech models. We protect your privacy by changing the pitch of your voice when voice commands are reviewed. All of our employees and vendors must also comply with strict privacy and security requirements in their handling of this information.
Do you use my Voice Commands for ads?
We do not use the content of your voice interactions to target ads. This means we do not use the content of voice commands such as “Hey Facebook, show me who’s online” as well as other sounds the microphone may pick up in the immediate area beyond your voice, like ambient noise or nearby background conversations.
Some of the resulting actions you take from Voice Commands, like the fact that you opened an app or played a game, may be used to inform the ads you see across Facebook. You can always delete your voice commands in the View Your Information Tool in your Privacy Center or on your Oculus app.
Why do you have humans review my voice commands?
When storage of your voice commands is turned on, we also use both machine processes and trained reviewers to train our voice services so that we can more accurately respond to your voice commands. Making sure our voice services can understand and respond to a wide range of speech patterns, phrases, local dialects, and accents is important to us, and trained reviewers are particularly helpful in identifying voice commands that machines are likely to misinterpret.
For example, someone might say "Open Quill," but Voice Commands could interpret what they said as "Open Chrome." Trained reviewers are also useful in evaluating how well our voice services execute your voice command. For example, we can evaluate if we correctly turned down the volume on your device or pulled up the website you requested based on your command. This helps improve our voice services to respond more accurately over time.
How do you protect my privacy during human review?
We have a team of vetted, trained transcription vendors and select employees who assist in reviewing voice commands. All of our employees and vendors must comply with strict privacy and security requirements in their handling of this information. We use controlled and monitored systems to ensure that the vendors and employees who are reviewing voice commands access only the information required to complete voice services training tasks. And we protect your privacy by changing the pitch of your voice when voice commands are reviewed so it's difficult to identify you.
How long do you keep my voice commands?
If a user decides to store their transcripts and related data to improve Voice Commands, their transcripts and related data about their voice interactions will be stored on Facebook for up to three years. If a user decides to store their voice recordings to improve Voice Commands, the voice recordings are stored on Facebook for up to three years. When we delete voice commands, they will no longer appear in the View Your Information Tool in your Privacy Center or on your Oculus app, will be deleted from our systems, and will no longer be used as part of our human review or machine learning processes. You can also always delete your voice commands in the View Your Information Tool in your Privacy Center or on your Oculus app. Alternatively, you can turn off Storage of your voice commands in Settings.
How do I use voice dictation in Oculus?
Tap on the microphone on your Oculus keyboard to use voice dictation instead of typing. For example, you can open Oculus Browser, tap the voice dictation button on the Oculus keyboard, and say “Oculus.com” to type Oculus.com into the URL field in Oculus browser.
What information do you collect when I use voice dictation?
When you use Voice Dictation, our system creates a short recording and transcription to process. Once we complete the dictation (that is, transcribe your voice to text), we delete the voice recording and transcription immediately, unless you choose to provide feedback on your Voice Dictation experience or have Oculus Dictation improvements turned on.
If you opt-in to improve Oculus Dictation, we store voice dictation data on your device, and only aggregated modeling information will be shared with Facebook to improve dictation for everyone.
To manage your Oculus Dictation improvements settings:
From your headset, select Settings then select Device.
Select System Keyboard and turn on or turn off Help improve Oculus Dictation.
How can I provide feedback on my voice dictation?
You have the opportunity to provide feedback on your voice dictation response by tapping the “thumbs-up” or “thumbs-down” icon after your voice dictation. If you tap the “thumbs-down”, you’ll have the option to submit the recording and transcription of your last voice dictation. If you choose to submit them, the recording and transcription will be stored on our servers for 90 days and a trained reviewer may listen for what went wrong to help make our voice services better for everyone. We protect your privacy during human review of the voice dictation you submit as described above in “How do you protect my privacy during human review”.
How does voice dictation differ from voice commands?
Voice commands and voice dictation are two separate services that allow you to use your voice to navigate Oculus. You can access voice commands by clicking the Voice Commands button in the Home menu. You can use voice dictation by pressing the microphone on the Oculus keyboard. We do not store or review recordings or transcripts of any voice dictations you make in the Oculus keyboard, unless you submit them to us for feedback or have Oculus Dictation improvements turned on. Because your voice dictation recordings and transcripts are not stored by default, you will not see them in your voice command logs.