How to record desktop audio on Windows

What you are actually trying to capture

"Desktop audio" (also called system audio, or "what you hear") is everything Windows plays out of your speakers or headphones: the game, the browser tab, the music app, the notification pings. Recording it sounds like it should be a checkbox, but Windows treats sound as something that flows out to a device, while recording apps only listen to things that flow in, like microphones. To record desktop audio, something has to loop the outgoing sound back around and present it as an input. Every method in this guide is a different way of building that loop.

The classic trick: Stereo Mix

For years the standard answer was Stereo Mix, a recording device some sound-card drivers expose that carries a copy of everything the card plays. To check whether you have it: open Settings > System > Sound, scroll to More sound settings, switch to the Recording tab, right-click the empty space, and tick Show Disabled Devices. If Stereo Mix appears, enable it, set it as your recorder's input, and you can record system audio today. That part is plain Windows and works right now, no extra software involved.

The catch is the word if. Stereo Mix is a feature of the audio driver, not of Windows itself, and modern drivers often skip it. USB headsets, HDMI audio through a monitor, and plenty of recent laptop chips simply do not offer it, and there is no supported way to add it back. If the device is not in that list after showing disabled devices, that driver does not have it, and no amount of searching will change that.

The other classic: recorders that capture it themselves

The second common answer is to use a recording app with system-audio capture built in. Screen recorders, the Windows game capture overlay, and some audio editors can grab desktop sound on their own using a loopback trick internally. If you just need one quick capture and the app you already use offers it, that is a perfectly fine answer, and it also works today.

What both approaches share: all or nothing

Stereo Mix and built-in capture have the same structural limit: they record the entire output as one baked blend. The email notification lands in the take. The message ping lands in the take. If a call rings while you are recording, that is in the take too. There is no mixing stage between "Windows played it" and "the recorder captured it," so there is nothing to turn down, nothing to leave out, and no way to keep your microphone on its own track. You get one stereo file of everything, and any cleanup happens after the fact, if it is possible at all.

The routing approach: a mixer between Windows and the recorder

A virtual audio mixer replaces the loop-everything-back trick with an actual mixing stage. In patchd, a Windows mixer currently in development, system audio already has a home: a fixed Desktop strip carries everything Windows plays, with its own fader and level meter, no setup required. Getting it to your recorder takes two moves:

• Route the Desktop strip to a bus. A bus is a virtual output that other apps see as a microphone. Each strip has a row of colored routing pills under a BUS header; click the Bus 1 pill on the Desktop strip and it lights cyan.
• Select the bus in your recording app. In your recorder's input picker, where you would normally choose a microphone, choose the patchd bus. The recorder now receives a clean, level-controlled feed of your desktop audio and captures it like any other input.

One posture note worth being precise about: patchd does not record anything itself. It routes and mixes, and your recording app, whether that is an audio editor, a DAW, or your capture software, does the actual recording. The division of labor is the feature: the mixer decides what is in the feed, the recorder just captures whatever the feed contains.

Want your voice in the recording too?

This is where the routing approach pulls away. Add your microphone as a strip, then decide how the recording should be shaped:

• One combined track: light the Bus 1 pill on both the Desktop strip and the mic strip. The bus now carries game plus voice as one mix, and your recorder captures it as a single input.
• Separate tracks: keep Desktop on Bus 1 and route the mic to Bus 2 instead. A recorder that accepts multiple inputs, a DAW or a multitrack recorder, records each bus as its own track, so you can edit the voice without touching the game. The same pattern drives separating game audio and voice chat, and it scales up to the multi-person layout in the podcast recording guide.

At launch, the free tier will include the fixed Desktop and Communications strips, your mic plus one virtual cable as selectable inputs, and three buses, which covers both layouts above with a bus to spare.

What stays out of the take

patchd builds mixes by inclusion: a source reaches a bus only when you light its pill there, so the recording contains exactly what you routed and nothing else. That fixes the two classic desktop-recording accidents. First, calls: chat apps live on the fixed Communications strip, separate from Desktop, so as long as its Bus 1 pill stays unlit, a mid-recording call reaches your ears but never the take. Second, single-app capture: the Desktop strip carries everything system-wide, so to record just one app, pull that app onto its own strip with a virtual cable and route only that strip to the bus. The browser tab you forgot about stays out because it was never routed in.

Where patchd is today

The Windows-native advice above is usable right now: check for Stereo Mix, and lean on your recorder's built-in capture if it has one. patchd itself is in development, so the routed path is what the app will do at launch rather than something you can install today. If a Desktop strip, colored bus routing, and a recorder that just picks a bus as its input sounds like the way this should have always worked, join the waitlist and we will tell you the moment it is ready.