Voice Control

Talk to Pro while you fly. Turn the runway rumble down on rollout, ask why you don't feel the stall buffet, or ask how hard that landing was, without letting go of the stick.

What it does

You hold a key, say one request and let go. Pro shows what it heard, makes the change or answers, and keeps every change in a draft you can compare with your original settings.

You can change forces, ask what a setting is set to or what it does, get help from the app's manual, and read values from the current flight, such as altitude or your last touchdown.

Commands are English only, whatever language the app is set to. Recognition runs on your computer after a one-time download of the voice models, and works offline after that.

Set up Voice Control

1. Click Voice Control in the sidebar.
2. Turn on the switch next to Voice setup.
3. Click Download voice models. The download window always includes the two models you need: the recognizer model, which turns your speech into text, and the understanding model, which works out what you meant. Select the natural voice or the wake word model as well if you want them, then click Download.
4. On the Input tab, choose your headset or desk microphone under Microphone. On macOS, allow microphone access when macOS asks.
5. Click Test microphone and speak. The INPUT meter should move with your voice. Click Stop microphone test.
6. On the Listening tab, look at Talk key. The keyboard key is F9. To talk with a button on your stick or yoke, click Bind next to JOYSTICK and press the button.
7. Type turn down runway rumble in Type a command or a question and click Run. When the change appears in the conversation, hold the talk key and say the same thing.

Whether the talk key works while the simulator is in front depends on your system:

• Windows: it works whichever window is in front.
• macOS: give Pro the Input Monitoring permission in System Settings. Without it, the key works only while Pro has focus.
• Linux: your desktop asks you to approve the talk key as a global shortcut. If you decline, or your desktop has no shortcut dialog, the key works only while Pro has focus. Click Bind again to retry.

A talk button on your stick or yoke works on every system, whichever window is in front.

Listening modes

Choose how Pro listens under Listening mode on the Listening tab.

Mode	How it works
Hold to talk	The default. Hold the talk key, say your request and release. Pro acts when you let go.
Tap to talk	Tap the key and speak. Pro stops listening when you stop talking or tap again. A long press works like Hold.
Always on, with the wake word	Pro listens all the time and starts a request when you say “Hey Bridge”. Download the wake word model on the Models tab first.

Require wake phrase is off by default, because pressing the talk key already tells Pro you are speaking to it. Turn it on if you want a second check: Pro then ignores any spoken request that does not start with Bridge, Tune or FFB, as in “Bridge, turn down runway rumble”. The option has no effect in Always on mode, and typed requests never need it.

Talking to Pro

Hold the talk key, say one request and release. You can also press and hold the Hold to talk button with the mouse. To check your wording without speaking, type the request in Type a command or a question and click Run.

Each exchange in Conversation shows:

• HEARD: the words Pro recognized
• CHANGES: each setting it changed, with the old and new value
• ANSWER: the reply, with any steps, tables and manual links below it

Open Recognition details to see how Pro understood the request. Copy puts the exchange on the clipboard, and Show earlier brings back older exchanges.

Change one setting per request. Pro remembers the last change for a short while, so “a little more” or “halfway between those” continue it, and “I meant engine rumble” corrects it.

Some requests wait for your approval: saving a profile, trying another profile, and changes to autopilot follow, the watchdog or the control system. Click Confirm command or Cancel, or say “yes” or “cancel”. If Pro is unsure what it heard, it asks you to say the request again instead of guessing.

Examples and help shows sample requests. Click one to put it in the typed box.

Original A and draft B

Every change you make by voice goes into a draft called B. Your settings from before the first change stay as A. You feel B on the stick as soon as a change is made.

Tuning comparison lists each changed setting with its A value, its B value and the difference. Use the controls under it to decide what to keep:

Control	What it does
Original A and Draft B	Switch the stick between the two, so you can fly the same approach with each.
Keep as A	Makes B your current settings and clears the draft. Save the profile on the Tuning page to keep them for the next flight.
Save draft as profile…	Asks for a name, such as “C172 soft ride”, and saves B as a new profile when you click Save.
Discard B	Throws the draft away and returns the stick to A.
Undo	Takes back the last voice change. Saying “undo that” does the same.

Saying “undo everything we changed” takes back up to ten voice changes from the last two minutes, after you confirm.

What you can say

Use the setting names from the Tuning page: “runway rumble”, “stall buffet”, “spring strength” and so on.

Changing forces

Say	What happens
“turn down runway rumble”	Lowers runway rumble one step.
“engine rumble up a little”	Raises engine rumble a small amount.
“set engine rumble to 15 percent”	Sets an exact value.
“reduce turbulence by 10 percent”	Lowers turbulence by that amount.
“turn off nosewheel shimmy”	Sets the effect to 0%.

Quick fixes

These change a group of related settings at once.

Say	What happens
“too much vibration”	Turns down ground, buffet, turbulence and engine vibration. The main stick force stays the same.
“taxi rumble is too strong”	Calms ground rumble, brake shudder, gear bumps, nosewheel shimmy and the taxi acceleration cue.
“trim feels wrong”	Turns trim relief back on, at full strength on both axes.
“too harsh”	Softens centering, airspeed load, the output ceiling and the strongest vibrations.
“too weak”	Raises the main stick forces and the strength of the effects you feel.

Questions

Say	What happens
“what is engine rumble set to”	Reads the current value.
“what makes it shake”	Names the forces you are feeling most right now.
“why don't I feel the stall buffet”	Explains why an effect is not playing.
“what does stick drop do”	Explains a setting.
“how do I fix a stick that does not move”	Answers from the manual and links to the section.
“what changed since takeoff”	Lists the settings changed during this flight.
“what's my altitude” or “how hard was that landing”	Reads the value from the simulator. If the simulator does not report it, Pro says so.
“open Black Box”	Opens that page.

A, B and profiles

Say	What happens
“switch to A” or “switch to B”	Same as the Original A and Draft B buttons.
“keep B”	Same as Keep as A.
“discard B”	Same as Discard B.
“save as soft ride”	Saves your settings as a new profile called “soft ride”, after you confirm.
“save profile”	Saves your settings to the current profile, after you confirm. Profile History keeps the version it replaced.
“try the 737 profile”	Loads that profile into B so you can compare it with A, after you confirm.
“remember this for this aircraft”	Saves a copy of your settings and uses it whenever you fly this aircraft, after you confirm.
“mark that”	Adds a marker to the flight Black Box is recording.

Safety

Say	What happens
“stop forces”, “forces off” or “panic”	Disarms the stick at once. Pro accepts these without the wake phrase.

Tune it for me

Say “help me tune engine rumble”, or click Tune it for me under Examples and help. Pro sets two values as A and B, you answer “A” or “B”, and it narrows the choice over a few rounds. It runs in Mock Sim or Flight Check, not during a live flight.

Requests Pro refuses

Voice Control does not do things that are hard to undo or that work around a safety limit. Pro replies that the request is blocked. Do these in the app instead:

• deleting or renaming a profile
• changing the hardware mode or the talk key
• downloading models, changing your license, exporting diagnostics or moving app data
• sending raw force commands, turning off safety limits, or “max everything”

Spoken replies

Spoken replies are off by default, and every answer appears on the page either way. To hear them, open the Replies tab and turn on Spoken replies.

Control	What it does
Voice	System voice uses your computer's own voice. For a more natural voice, click Download natural voices and choose one.
Speaking speed and Volume	Set how fast and how loud Pro speaks.
Preview voice	Speaks a sample reply. Stop Preview ends it.
Confirmation tones	A short tone confirms a command or asks you to repeat it. On by default.

For a long answer, Pro speaks the first part and leaves the rest on the page. Click Stop speaking or press Esc to cut a reply short.

Phrases and Diagnostics

The Phrases tab holds two lists:

• Personal voice shortcuts: your own phrase for a command. For example, save “bumpy day” as “reduce turbulence by 10 percent” and click Save shortcut. Pro asks for confirmation every time you use one.
• Corrections: words Pro keeps mishearing. Enter the Misheard words and the Intended words, then click Add correction. Saying “I said engine rumble” after a wrong result adds one too.

The Diagnostics tab shows which part of the chain is working: the microphone, the recognizer, the understanding model, the reply voice and the talk key. CONFIDENCE shows how sure Pro was of the last spoken request. Below 0.50 it does not act and asks you to repeat.

Store unresolved transcripts locally is a separate setting, off by default. When it is on, Pro keeps the text of requests it could not handle, so you can fix the words with Correct words or copy them into a report. It never stores audio.

Privacy

Pro processes your voice on your computer. Audio is never uploaded. Pro goes online for voice only when you download or update the models.

To report a voice problem, click Report an Issue in the sidebar. The report contains text, never recordings, so write what you said, what Pro heard and what you expected.

Safety

When something goes wrong

Problem	What to do
The talk button reads Need model	Open the Models tab and click Download voice models. Wait until each model reads Ready. A cancelled download continues where it stopped.
The INPUT meter does not move	Check the microphone chosen under Microphone and that it is not muted. Click Rescan microphones if you plugged it in after opening the page. Allow microphone access for Pro in Windows Settings or macOS System Settings under Privacy & Security, then reopen Pro.
The meter moves but nothing happens	Hold the talk key until you finish speaking. If Require wake phrase is on, start with Bridge, Tune or FFB. If the key works only while Pro is in front, see the talk key notes in Set up Voice Control.
A typed command fails too	Check that Voice Control is switched on. Name one setting per request, using its name from the Tuning page, or try a request from Examples and help.
Pro ignores a command or changes the wrong setting	Open Recognition details to see what Pro understood. Say “undo that”, then add a correction on the Phrases tab for the misheard word.
Replies are silent	Turn on Spoken replies, check Volume and your computer's sound output, and click Preview voice. On Linux the system voice needs Speech Dispatcher with a working voice; the natural voice works without it.

For other problems, see Pro troubleshooting.