Two different audio streams
- Microphone input - your voice. Available to any application with permission, on every platform.
- System output - everything the device plays, including the other person's voice. Access to this is restricted, and the restriction is the whole problem.
A tool that advertises transcription but only requests microphone access can, by construction, transcribe only you. For interview use that is the half that matters least, since the questions come from the other side.
How it works per platform
| Platform | Capturing output audio | Practical note |
|---|---|---|
| Windows | Supported natively via loopback capture | Generally works without extra setup |
| macOS | No native loopback; needs a driver or the newer system capture APIs | Historically required a virtual audio device |
| Browser extension | Only the audio of a tab it is permitted to capture | Cannot hear audio outside that tab |
| Phone call on an actual phone | Not capturable by a desktop app at all | The audio never reaches the computer |
Why a native app and a browser extension differ here
An extension is scoped to the browser, and typically to a tab it has been granted capture rights on. If the interview runs in a desktop Zoom or Teams client rather than a browser tab, an extension has nothing to listen to. A native application can request system-level output capture and is therefore indifferent to where the call is running - which is the main structural reason desktop tools cover more cases. The mechanics are covered in how AI interview assistants capture system audio and the comparison in Chrome extension vs desktop.
What to verify before you rely on it
- Does it ask for system or output audio permission, not just microphone? If only microphone, it hears you alone.
- On macOS, does it need a separate audio driver, and are you willing to install one?
- Does it work with the desktop client of your meeting tool, or only a browser tab?
- Test with a real call before the interview. A recorded test where you can confirm both voices appear in the transcript is the only reliable check.
- What happens with headphones - some capture paths behave differently when output is routed to a headset.
The consent question
Recording a conversation is governed by law that varies by jurisdiction, and in some places all parties must consent. Transcribing for your own live assistance is not the same as retaining a recording, but the distinction is not uniform either. If you intend to keep a recording, check the rule where you and the other party are, and when in doubt ask - an interviewer who is asked usually says yes, and one who discovers an undisclosed recording usually does not stay comfortable.
Where CoPilot Interview fits
Disclosure: we build an interview assistant, so treat this as a disclosed pitch. CoPilot Interview is a native Windows and macOS app that puts suggested answers on your own screen in about two seconds, across coding, system design and behavioral rounds, with a permanent free tier that needs no credit card.
Frequently asked questions
Can an app record both sides of a phone interview?
Only if it captures the device's output audio as well as the microphone, and only if the call actually runs through that device. A desktop app cannot capture a call made on a mobile phone, because that audio never reaches the computer.
Why does my transcription only capture my voice?
Because the tool has microphone permission but not system or output audio access. The microphone records only your side; the other person's voice arrives through the device's output, which requires a separate and more restricted capture path.
Is capturing system audio harder on macOS than Windows?
Yes. Windows supports loopback capture of output audio natively, while macOS has historically required a virtual audio driver or the newer system capture APIs. That is why some tools work out of the box on Windows but ask Mac users to install an audio component.
Do you need consent to record an interview?
It depends on jurisdiction, and some require all parties to consent. Live transcription for your own assistance is not identical to retaining a recording, but the legal line is not uniform. If you intend to keep a recording, check the local rule and consider simply asking - most interviewers agree when asked in advance.
Walk into the round with a second pair of eyes
CoPilot Interview is a native Windows and macOS app that puts suggested answers on your own screen in about two seconds, across coding, system design and behavioral rounds. The free tier is permanent: 15 minutes a day, no credit card.
Try it free