Skip to main content
Voice command aliases define the phrases that ORbit’s voice system recognizes in Room Mode. Each alias maps a spoken phrase to a specific action — recording a milestone, canceling one, undoing the last action, or calling the next patient.
Voice alias management is done in the web app only. The iPad Room Mode consumes aliases but doesn’t provide a settings UI for managing them.

How aliases work

When someone speaks a command in Room Mode, ORbit checks the alias dictionary for an exact match. If found, the command executes instantly with no network call. If no match is found, the command is sent to an AI classifier for interpretation.

Alias types

Milestone aliases

Map a phrase to a specific milestone action:
FieldDescription
PhraseThe spoken text to match (e.g., “patient in room”)
MilestoneThe facility milestone this maps to
Actionrecord (stamp timestamp) or cancel (clear timestamp)

General command aliases

Map a phrase to a system action (no milestone association):
Action typePurposeExample phrases
undo_lastUndo the most recent voice-recorded milestone”undo”, “undo that”, “take that back”
next_patientTrigger call-next-patient workflow”next patient”, “call the next patient”
surgeon_leftMark surgeon as departed”surgeon left”, “the surgeon has left”
confirm_pendingConfirm a pending out-of-order command”confirm”, “yes”, “go ahead”
cancel_pendingCancel a pending command”cancel”, “no”, “never mind”
start_cement_timerStart the cement timer overlay”start a cement timer”
stop_cement_timerStop and dismiss the cement timer”stop the cement timer”

Managing aliases

Viewing aliases

Navigate to Settings → Voice Commands to see all aliases for your facility. The table shows:
  • Phrase — the spoken text
  • Milestone — which milestone it maps to (blank for general commands)
  • Action — record, cancel, or general command type
  • Source — manually created or auto-learned
  • Status — active or inactive

Adding aliases

  1. Click Add Alias
  2. Select the action type (record, cancel, or a general command)
  3. For milestone actions, select the target milestone
  4. Enter the phrase — this should be lowercase, without punctuation, and without the “Orbit” wake word
  5. Save
Add multiple natural variations for important milestones. For example, “Patient In” might have aliases: “patient in”, “patient in room”, “patient is here”, “patient arrived”.

Deleting aliases

Click the delete button on any alias row. Auto-learned aliases can be freely deleted — they’ll be re-learned if the same phrase is spoken again with high AI confidence.
Global seed aliases (those shared across all facilities) cannot be deleted by facility admins. Contact a global admin to modify seed aliases.

Auto-learning

When ORbit’s AI classifier recognizes a new phrase with high confidence (85%+), it automatically:
  1. Saves the phrase as a new alias tagged auto_learned = true
  2. Strips the “Orbit” wake word prefix (so “Orbit, the patient is ready” becomes “the patient is ready”)
  3. Scopes it to the current facility
  4. Adds it to the local dictionary immediately — the next use is an instant match
Auto-learned aliases appear in the settings table with an “Auto-learned” badge. You can review and delete them at any time.

What triggers auto-learning

ConditionAuto-learns?
AI confidence ≥ 85% and shouldCache = trueYes
AI confidence 50-84% (user confirmed)No — confirmation doesn’t trigger auto-learn
AI confidence < 50% (rejected)No
Phrase already exists as an aliasNo — deduplication prevents duplicates

Facility scoping

All aliases are scoped to a facility. Each facility can have its own set of aliases reflecting local terminology and workflow preferences. When a user opens Room Mode, only aliases for the current facility are loaded.

Best practices

1

Start with seed aliases

Every new facility gets a set of default aliases from the milestone template. These cover the most common phrases for each milestone.
2

Let auto-learning do the work

Use voice commands naturally for the first few days. ORbit will auto-learn your team’s preferred phrasing. Review the auto-learned aliases in settings and delete any that don’t make sense.
3

Add common variations manually

If your team uses specific terminology (e.g., “knife to skin” for incision), add it manually so it matches on the first try.
4

Clean up periodically

Review auto-learned aliases monthly. Remove phrases that were one-time misrecognitions or that don’t represent reusable commands.

FAQ

Yes. Aliases are stored in the Supabase database and sync across all iPads at the same facility. An alias added on one device is available on all devices.
No. All phrases are normalized to lowercase before matching. “Patient In”, “patient in”, and “PATIENT IN” all match the same alias.
No. Each phrase must be unique within a facility. The system enforces this with a database constraint to prevent ambiguous matches.
The phrase will no longer match instantly. If someone says it again, it will go through AI classification and may be auto-learned again if the AI classifies it with high confidence.

Next steps

Voice commands guide

Learn how to use voice commands in Room Mode.

Milestone settings

Configure milestone templates and ordering.