Two opt-ins on a new Audio screen, both ported from Bannerlator's gamescope runtime where they are proven with voice in Steam's own tester. DirectAudio replaces the audio driver inside the games the client launches - Wine's mmdevapi driver, glibc build - with one that talks to a helper on the Android side over a unix socket. The client keeps PulseAudio either way: it is a native Linux program with no other way to make a sound. The helper runs here because nothing inside the session can open Android's audio: a game is a glibc process in a rootfs that binds no /system, and the uid the helper runs under is what the recording permission is checked against. The microphone is its own opt-in. The helper opens the input stream only when asked and writes it to a pipe; PulseAudio's module-pipe-source turns that into a source the client sees as DirectAudioMic. The daemon makes the pipe and the helper waits for it - the two used to race, and the module fails on EEXIST. The Proton wrappers gain Bannerlator's bl_directaudio, which adds the driver's directory to WINEDLLPATH and names the driver in the prefix, and refuses on any Proton that is not Wine 11 rather than pair a driver with an interface it was not built against, which would be silence rather than an error. bl_game_cpus comes along in the same splice and is inert until BL_GAME_CPUS is set. The driver is staged into the runtime at every launch like the session scripts; the helper ships as libdirectaudiorelay.so beside the PulseAudio daemon. Both are the binaries Bannerlator carries, built by The412Banner/directaudio's relay workflow (linux-wine11 build against ValveSoftware/wine proton_11.0).
DirectAudio for the Linux Steam client (relay build)
This is the driver for games that run under Valve's ARM64 Proton inside a Linux runtime on Android - Bannerlator's linuxfs session with the native Steam client - where the game process is a glibc process and cannot touch Android's audio libraries itself.
Two pieces, both required:
| piece | what it is | where it runs |
|---|---|---|
aarch64-unix/winedirectaudio.so + the two PE shells |
the Wine mmdevapi driver, glibc build, no AAudio inside |
in the game, inside the Linux rootfs |
directaudio-relay |
a small bionic program that owns the AAudio output and microphone streams | on the Android side, under the app's uid |
They meet over a unix socket. The driver connects, sends its launch config, and gets back two shared-memory rings: one it fills with mixed game audio, one the helper fills with microphone audio. The helper plays and records; the driver keeps doing everything else it always did.
Which Proton
Built inside ValveSoftware/wine at proton_11.0. The private mmdevapi
interface it implements is byte-identical across Proton 11.0 (ARM64),
Proton Experimental (ARM64) and our own Wine-11 layers (checked 2026-09-19),
so this one build serves any of them. A future Proton on a Wine 12 base will
need a rebuild - the interface is not versioned, and a mismatch is silence, not
an error.
Installing the driver next to Proton (not into it)
Steam verifies and updates its Proton depots, so files added under
steamapps/common/Proton .../files/lib/wine/ will not survive. Put the driver
in a directory of its own and point Wine at it:
<somewhere>/directaudio/lib/wine/aarch64-unix/winedirectaudio.so
<somewhere>/directaudio/lib/wine/aarch64-windows/winedirectaudio.drv
<somewhere>/directaudio/lib/wine/i386-windows/winedirectaudio.drv
and set, for the game process:
WINEDLLPATH=<somewhere>/directaudio/lib/wine
Wine searches WINEDLLPATH entries with the per-arch subdirectories appended,
so both the PE shell and the unixlib are found there. In Bannerlator's runtime
the natural place to export it is the bannerlator-proton compatibility-tool
wrapper, which every Proton launch goes through.
Selecting the driver
mmdevapi only tries the drivers named in the registry (or its built-in
pulse,alsa,oss,coreaudio list, which does not know about this one). Set, in the
game's prefix:
wine reg add "HKCU\Software\Wine\Drivers" /v Audio /d directaudio /f
Under Proton the prefix is compatdata/<appid>/pfx; run the command with that
prefix as WINEPREFIX using Proton's own bin-arm64/wine, or write the key from
the wrapper before it hands off to Proton.
Running the helper
directaudio-relay --socket <path> [--log]
- Run it under the app's uid, the way the app already runs its PulseAudio
daemon. Android checks
RECORD_AUDIOagainst the calling uid, so a helper started any other way can play but cannot record. - Start it before the game: the driver asks "is a helper there?" once, when Wine picks an audio driver, and falls back to unavailable if not.
- The socket path must be reachable from inside the Linux rootfs. Bannerlator binds the app's files directory and its runtime directory into the session at their own paths, so any socket under either works unchanged on both sides.
- The helper serves any number of game processes; each gets its own streams.
It exits on
SIGTERM; a game exiting tears down that game's streams only.
Telling the driver where the helper is
BANNER_AUDIO_DIRECT_RELAY=<the same socket path>
Unset, the driver looks for $XDG_RUNTIME_DIR/directaudio-relay.
Everything else is unchanged
All BANNER_AUDIO_DIRECT_* knobs mean what they mean in the in-process build:
_PERF, _ADAPTIVE, _DECAY, _MS, _MAXMS, _BF, _MBF, _EXCLUSIVE,
_WATCHDOG, _STALL_MS, _DECAY_*, _LOG, _PERIOD_MS, _MINPERIOD_MS, and
the live-config mailbox _RUNTIME (the helper watches the file; it must be a
path the helper can read on the Android side). BANNER_AUDIO_DIRECT_MIC=1
exposes the capture endpoint exactly as before: the helper opens the input
stream (VOICE_COMMUNICATION preset) when the game connects and makes it hot on
the game's first capture start.
One thing is different: there is one extra hand-off, the ring. The helper's
callback pulls a burst from it, so the driver keeps two bursts queued (about 8 ms
on a 192-frame device) on top of the AAudio buffer the presets control, and the
helper raises that if a callback ever finds the ring short. get_latency reports
the sum, so a game that sizes its own buffers from it still sees the truth.
Proving it is running
- Helper side:
logcat -s DA-Relay:Ishowshello from <exe>, thenopen: buffer ... ring target ..., thencapture open/capture startfor the mic. - Game side: the session log carries
DirectAudio: relay: connected - .... - The game process's
/proc/<pid>/mapsshowswinedirectaudio.soand nolibaaudio.so(it is in the helper's maps instead). The AudioFlinger track is owned by the helper's pid.