Confidant records from whatever microphone you've selected. Here's how to pick the right one and how to capture audio from a video call.
Picking your input device
Open Settings → Recording → Audio Input
Click the device dropdown
Pick the device you want — built-in mic, USB headset, virtual device, etc.
If you regularly switch between mics or headsets, picking a specific device explicitly avoids surprises.
The first time you actually start a recording (not when you select the device), macOS or Windows will prompt for microphone permission. Approve it.
Hardware that works well
This is about the microphone you record from — capturing remote audio for telehealth is a separate setup, covered below.
In rough order of how well they transcribe in-room speech:
Wired USB or analog headset — close to your mouth, isolated from room noise
External USB condenser mic (Yeti, Shure MV7) on a desk arm — great if you sit still
Built-in laptop mic — usable but picks up keyboard noise and room reverb
Wireless earbuds with mic (AirPods, similar) — works, but the mic is noticeably worse than wired and battery life can become an issue mid-session
Background noise of any kind shows up in the transcript. A quiet room with a wired mic produces dramatically better notes than a coffee shop with AirPods.
Capturing audio from a video call (telehealth)
If your sessions are over Zoom, Doxy, or another telehealth tool, you'll need to route the call's audio into Confidant. Whether or not you need a virtual audio device depends on your setup — some Macs already pass system audio through their default input, in which case you don't need to install anything extra. Try recording a short test session first; if your client's audio is missing from the transcript, set up a virtual audio device using the steps below.
macOS — BlackHole setup
What is BlackHole? A free, open-source helper for macOS (made by a third party — not part of Confidant) that creates a virtual audio "input" device. Without it, your microphone only captures what's spoken in your room — your client's voice plays out of your speakers but the computer has no way to record it. Once installed, BlackHole shows up alongside your other input devices, and Confidant can record from it like any other mic.
For convenience, Settings → Recording → Audio Input in Confidant has a "BlackHole" link that expands an in-app guide with the same steps below:
Install BlackHole 2ch (free, open source)
Open Audio MIDI Setup (in /Applications/Utilities)
Click + → Create Multi-Output Device and check both your speakers and BlackHole 2ch
Set the Multi-Output Device as your system output in System Settings → Sound → Output
In your video call app, set the output to the Multi-Output Device (or your speakers)
Back in Settings → Recording → Audio Input, select BlackHole 2ch as the recording device
Confidant then captures whatever the call is playing while you still hear it through your speakers.
What this captures
The BlackHole route captures the remote voice (your client) cleanly. To capture your own voice as well, you have two options:
Speak loudly enough that BlackHole picks up your voice through your speakers (works for many setups, but your own voice will sound thinner in the recording)
Set up an Aggregate Device that combines BlackHole + your physical microphone, then pick the Aggregate Device in Settings → Recording → Audio Input. macOS's Audio MIDI Setup creates Aggregate Devices the same way it creates Multi-Output Devices.
If unsure, start with the simple BlackHole setup and check the transcript after a test call. If your voice is missing, switch to the Aggregate Device approach.
Windows
Use VB-Audio Virtual Cable (free) — the equivalent of BlackHole on Windows. Same principle: route your call audio into the virtual cable, then either select that cable as your input in Confidant, or aggregate it with your physical mic before recording.
FaceTime + earphones — not supported
If you do telehealth over FaceTime while wearing AirPods, Bluetooth headphones, or any other earphones, your client's voice will not be captured. FaceTime on macOS routes audio through a low-level path that bypasses the standard system audio stream Confidant captures from. Other call apps (Zoom, Doxy, Google Meet, etc.) don't have this issue because they use the regular audio output that Confidant can hear.
If you use FaceTime for sessions, choose one of these alternatives so the recording captures both voices:
Use your laptop's built-in speakers instead of earphones, or
Switch to Zoom / Doxy / another platform when you want to wear earphones, or
Set up the BlackHole or Aggregate Device route described above (which routes FaceTime through the captured path)
Verifying it's working
Start a session. The session panel shows a live audio level meter. If you talk and the meter doesn't move, the wrong device is selected (or the mic is muted at the OS level).
Stop and restart the session after switching devices — Confidant locks in the device choice when recording starts.