diff --git a/packaging/void/srcpkgs/voculinux/INSTALL.msg b/packaging/void/srcpkgs/voculinux/INSTALL.msg
index fead780..308e3aa 100644
--- a/packaging/void/srcpkgs/voculinux/INSTALL.msg
+++ b/packaging/void/srcpkgs/voculinux/INSTALL.msg
@@ -9,6 +9,11 @@ Speech models are not in this package: they
 download on first run into ~/.local/share/voculinux
 and are checked against pinned checksums.
 
+Audio capture uses PipeWire (pw-record and pw-dump,
+from the pipewire package this one depends on), so
+PipeWire must be running in your session. PortAudio
+(python3-PyAudio) is the other capture path.
+
 Optional engines are manual additions outside the
 guaranteed set: vosk, OpenAI whisper, Parakeet
 (sherpa-onnx), faster-whisper and onnxruntime (VAD)
@@ -26,4 +31,4 @@ injection fallback rung is absent.
 
 Update through your package manager instead of the
 in-app release downloads:
-  sudo xbps-install -Syu voculinux
+  sudo xbps-install -Su voculinux