# Sound for creators

> Recording, streaming and producing: buffers, OBS, a separate game track, DAWs.

Source: https://wolos.pages.dev/manual/creators-audio

## What it is

What works out of the box for people who record, stream or produce, what one command adds, and the three things to know before a session.

## Why

Creators need two things from sound that everyday users do not: tiny buffers while monitoring, and separate tracks for the game and the voice. Both are choices, so they are switches, not defaults.

## How

Run `wol audio doctor` first. Pick the buffer for the job in Settings > Sound > Pro Audio, and use `wol audio appmix on` when a game and a voice need their own tracks.

## If something goes wrong

| What you see | What to do |
|---|---|
| The microphone clips | Run `wol audio doctor`; it names the input level and the fix. Keep the slider at or below 100% unless you need *Allow Louder*. |
| Dropouts while monitoring | Raise the buffer in Settings > Sound > Pro Audio (64 frames for monitoring through a DAW, 256 for a stream). |
| OBS does not see the game audio | Run `wol audio appmix on`, send the game to *WolOS App Mix*, and capture it in OBS. `wol audio appmix status` says which pieces exist. |
| Flatpak OBS cannot reach sound | `flatpak override --user --filesystem=xdg-run/pipewire-0 com.obsproject.Studio`; the native package is simpler. |

## Related

[Sound](/manual/sound) · [Optional modules (Creative)](/manual/modules)

## Reference

What works out of the box on WolOS, what one command adds, and the three things to know before a session. The rules behind it are in `modules/audio.md`.

### Before you record

1. **100% is 0 dB.** The input slider's top is the microphone's nominal level, not the hardware maximum. If a mic is quiet at 100%, turn on *Allow louder* in Settings > Sound > Input and watch the meter; red means clipping.
2. **Check the stack:** `wol audio doctor`. It names the default input's level in dB, dropouts, forced buffers and the OBS pieces, each with the fix.
3. **Pick the buffer for the job** in Settings > Sound > Pro Audio: 64 frames (1.3 ms) for monitoring through a DAW, 256 for a stream, automatic for everything else. The choice is written to `~/.config/pipewire/pipewire.conf.d/50-wolos-pro-audio.conf` and applied live; it survives a reboot and `wol audio reset` removes it.

### OBS Studio

Install OBS (`pacman -S obs-studio`) and the PipeWire audio capture plugin (`pacman -S obs-pipewire-audio-capture`, from the [wolos] repo). The plugin adds three sources: **Application Audio Capture** (one app, picked by name), **Desktop Audio Capture** (what the speakers play) and **Audio Input Capture (PipeWire)**. Screen and window capture use the compositor's portal (`xdg-desktop-portal-hyprland`): OBS asks, you pick.

**Game audio and voice on separate tracks.** `wol audio appmix on` creates an output called **WolOS App Mix** and a source called **Mic + App Mix**:

- send the game or the browser to *WolOS App Mix* (Settings > Sound > Apps, or the app's own output picker); you still hear it on your speakers or headset;
- in OBS add *Application Audio Capture* on *WolOS App Mix* for the game track and *Audio Input Capture* on your microphone for the voice track, or one *Audio Input Capture* on *Mic + App Mix* when a single mixed track is enough.

`wol audio appmix off` removes both; `wol audio appmix status` says which exists. They are pipewire-pulse modules, so they go away with a logout too.

**Flatpak OBS** needs the PipeWire socket: `flatpak override --user --filesystem=xdg-run/pipewire-0 com.obsproject.Studio`. The native package is simpler.

### Voice: automatic level and noise

Settings > Sound > Input has two switches. *Automatic input level* puts WebRTC gain control, noise suppression and echo cancel in front of the microphone; the result is a new input, **WolOS Microphone**, which becomes the default so every app uses it. *Reduce background noise* adds the RNNoise filter (`noise-suppression-for-voice`, on the image) and shows up as **WolOS Microphone (noise reduced)**. Both are loaded only while switched on. For OBS, pick the processed input as your microphone source.

### DAWs and JACK apps

`pipewire-jack` is the JACK on the image: Ardour, Reaper, Bitwig, Carla and friends see PipeWire as a JACK server with the buffer you chose in Pro Audio. Realtime scheduling comes from rtkit, and members of the `audio` group have `rtprio 95` / `memlock unlimited` directly (`/etc/security/limits.d/70-wolos-audio.conf`). A USB interface keeps its hardware mixer (preamp gain in the interface's own knob), and its card can be switched to the *Pro Audio* profile in Settings > Sound > Pro Audio, which exposes every channel raw and bypasses the automatic routing.

Sample rate: the graph runs at 48 kHz but follows 44.1, 88.2 and 96 kHz streams instead of resampling them; force one in Pro Audio for a whole session. Dropouts show as *xruns* in Pro Audio and in `wol audio doctor`; the fix is a larger buffer, another USB port, or a powered hub.

### Bluetooth headsets

The best codec the headset offers wins (LDAC, aptX HD, aptX, AAC, SBC-XQ). Recording through a Bluetooth headset switches it to the call profile (HFP, lower quality) only while an app records, and back afterwards; turn *Switch to headset mode for calls* off in Settings > Sound if you prefer to keep music quality and use another microphone. The codec can be chosen per device.

### When something is wrong

- `wol audio doctor`: the checks and the fixes.
- `wol audio reset`: shows what a reset would do; `wol audio reset --yes` forgets stored defaults and volumes, removes the Pro Audio file and restarts PipeWire. Your apps reconnect by themselves.
- Settings > Sound > Troubleshooting has the same reset, *Keep device awake* (for amplifiers that click when PipeWire suspends them) and *Ignore hardware volume* (software volume for a device whose mixer lies).