Your MacBook's Microphone Captures More Than Your Voice During Respondus Monitor (2026 Audit)
What the microphone actually captures
| Source | Recorded? | Notes |
|---|---|---|
| Your voice | Yes | Primary intended capture |
| Background voices (family, roommate, TV) | Yes | Picked up by built-in or external mic |
| macOS notification sounds (iMessage, email) | Yes | Routed to the same audio mix on most Macs |
| Music or video from another app | Mostly no - unless routing to mic input | Depends on Mac audio routing config |
| Bluetooth calls (FaceTime, phone) | Yes if active during exam | Audio is mixed at the OS level |
| Smart-home announcements (HomePod, Echo) | Yes - picked up acoustically | If audible to you, audible to mic |
| HVAC, appliance noise | Yes | Often misclassified by detector |
How macOS routes audio during an LDB session
By default, LockDown Browser receives audio from the input device selected in System Settings → Sound → Input. This is typically:
- Built-in microphone: MacBook lid mic. Picks up everything in the room.
- AirPods / Bluetooth headset mic: directional but still environmental.
- External USB mic: if you've set one as default.
The mic input is a single audio stream. macOS does not, by default, mix multiple sources into the mic - but third-party tools (BlackHole, Loopback, Audio Hijack) can do so. If you've installed any of these, audit your routing before exams.
What gets flagged for review
Respondus Monitor's audio detector flags:
- Speech beyond your voice (multiple speakers detected).
- Sustained sounds matching "conversation" patterns.
- Sudden volume spikes.
It does NOT flag:
- Single-voice background hum (HVAC, fan).
- Notification sounds (one-shot, short).
- Music with vocals (sometimes flagged as "speech", inconsistent).
The detector is heuristic. False positives are common in noisy environments. See dispute procedure if a flag is unfair.
How to harden the audio environment
- Notifications: Do Not Disturb. System Settings → Focus → Do Not Disturb → On. Mutes iMessage, email, app notifications during the exam window.
- Quit background apps. Music, video, podcast apps. Anything with audio output potential.
- Mute system sound. Sound is mostly output-side, but some apps play after sleep. Mute.
- Disconnect Bluetooth audio devices not in use. AirPods auto-connecting mid-exam is a known cause of audio glitches and "voice in the room" flags.
- Tell household members about the exam window. A written note on the door cuts mid-exam interruptions ~70%.
- Use a directional mic. A USB cardioid mic picks up your voice clearly while attenuating room sound - reducing false-flag triggers.
- Audit installed audio routing tools (Loopback, BlackHole, Audio Hijack). Confirm your Mac's default mic input is the physical microphone, not a virtual device.
Privacy implications worth flagging
- If a family member discusses unrelated personal matters during your exam, that audio is captured and stored on Respondus servers for the institution's retention period.
- Some institutions use the recordings for research, training models, or staff review - check your university's data-use disclosure.
- If you want to suppress the mic capture entirely, your only option is to disable the Microphone permission, which will prevent LDB from launching the exam.
Frequently asked questions
Will the microphone capture audio from my AirPods if they're connected to my iPhone, not my Mac?
No - audio routing is per-device. AirPods connected to your iPhone are not on your Mac's mic input path. However, if your iPhone is in earshot of your Mac's built-in mic, anything audible (FaceTime call playback, etc.) is recorded acoustically.
Can I use noise-cancelling headphones during the exam to reduce background noise?
Mostly - headphones suppress what YOU hear, not what the mic captures. To reduce mic capture of room noise, use a directional mic with noise-cancelling INPUT (e.g., Shure MV7+ or AirPods Pro 2 with their voice-isolation mode active).
Are my recordings used for training Respondus's detection models?
Per Respondus's public statements, recordings are used only for instructor review and not for model training without separate consent. Training data is sourced from synthetic or opt-in datasets. This is documented in their privacy policy - read the version current to your enrollment date.