Arctis Sound Manager has been released in version 1.4.25.

Following version 1.4.24, this update mainly fixes issues on the Nova Pro Omni family: the ChatMix knob often did nothing after boot because the base station aborted the mode switch when the next command followed too quickly - profiles can now insert deliberate delays into device initialization (contributed by @kevinpbaker). Also fixed: DeepFilterNet silently killed the microphone, a swapped Volume Limiter/Wear Sense pair on GameBuds, and a leak of GPS coordinates in bug reports. New additions include independent ANC intensity for GameBuds and four extra Sonar presets.

Fixed

• Nova Pro Omni: the ChatMix knob did nothing after boot until the daemon was restarted. The base station only reports its mixer to the PC once software has switched it into that mode (0x8d sonar-present, 0x49 software ChatMix). It accepts both commands and then abandons the switch if the next command lands within a few milliseconds — which it always did, since init sends its frames 6 ms apart. Pressing the knob showed no mixer, turning it only moved the station volume, and restarting the daemon by hand replayed the same too-fast burst. Profiles can now put ['sleep', <ms>] in device_init; the Omni waits a second after each mode switch. Verified on hardware, including from a cold base station. Contributed by @kevinpbaker. (#238, #235, #245)
• A failed init frame was never retried. send_command() logs USB errors and returns, it does not raise, so the “retry once” written around it in the init sequence could never fire. It now reports failure and the retry runs.
• ChatMix on the Nova Pro Omni also stayed dead specifically after a system resume, on top of the boot-time issue above. resume_from_sleep() used to replay device_init over the same libusb handle that had just survived suspend; on this family that left the ChatMix event stream dead until a physical replug even though ordinary commands kept working. configure_virtual_sinks() — the already-tested path that properly releases and re-acquires a same-device re-enumeration — is used instead. Checking SteelSeries’ own engine specification against ASM turned up two more protocol timings ASM never respected on any profile: a firmware-family minimum spacing between consecutive commands (time_between_commands_ms, 1-50ms depending on family) and a settle time the transmitter needs after enumerating before it can be talked to at all (init_sleep_length_ms, up to 5s on the Nova Pro Omni/Elite/Wireless/GameDAC family) — both are now honoured for every headset profile, sourced from SteelSeries’ own specifications. The device settle wait runs off the USB hotplug thread via a background timer so it can never delay detecting other USB devices. (#238)
• ASM never told the DAC it was leaving. SteelSeries’ own engine sends disable-ChatMix/disable-Sonar to the Nova Pro Omni/Elite/Wireless/Wired and Nova 3 Wireless families before releasing the USB interface, so the base station falls back to plain hardware volume instead of staying in software-ChatMix mode with nothing driving it once the host goes away. teardown() now does the same, best-effort, on every daemon stop or disconnect.
• After the wireless headset itself reconnects (powers back on, or comes back into range), ASM trusted the very first settings read/replay attempt instead of retrying. SteelSeries’ own engine retries a status read up to 10 times after a radio reconnect on several families, because the first attempt(s) right after reconnect can come back stale — ASM’s own settings replay now does the same before pushing anything back to the device, falling back to the old flat-delay behaviour on families where this was never observed to matter.
• DeepFilterNet noise cancellation silently stopped the microphone. generate_sonar_micro_conf() assumed every LADSPA node in the micro chain exposes ports named “Input”/“Output” — true for RNNoise, the noise gate and the compressor, but DeepFilterNet’s plugin names them “Audio In”/“Audio Out” instead. The generated filter-chain config referenced ports the loaded plugin doesn’t have: no audio path was built and the mic went silent with the node still “running”, no crash, nothing in the logs. Configs already broken by this are now auto-detected and repaired instead of staying stuck. (#240)
• A native (non-PulseAudio) application stream with no application.name property was invisible to the mixer, even though it was still playing and still moved by the Output volume slider. get_native_streams() only looked at that one property to identify a stream; a raw ALSA/native client that never sets it — reported case: SMAPI-launched Stardew Valley, a .NET/Mono app under Distrobox — was silently dropped instead of falling back to node.name/application.process.binary, so there was no way to drag it onto Game/Chat/Media. (#243)
• The external Output channel could get silently and permanently stuck pointing at nothing after a device was connected while ASM was already running (e.g. a TV switched on over HDMI). Resolving the new device’s sink can fail on the very first attempt because it hasn’t finished settling in the PipeWire graph yet; that empty result used to get locked in as “reconciled” regardless, so the watchdog never retried it even once the device was fully there. Only a full daemon restart used to force a fresh resolution. (#246)
• GameBuds: the Volume Limiter and Wear Sense toggles were silently swapped, each one controlling the other’s feature, since the profile’s opcodes didn’t match SteelSeries’ own specification (0x27 is actually the volume limiter, 0xc5 is actually wear sense).
• Bug reports no longer leak the weather widget’s raw GPS coordinates (weather_lat/weather_lon) — they were the one field the existing “location”/“city” redaction patterns didn’t match, while the city name next to them was correctly stripped.

Added

• GameBuds: ANC intensity, and independent playback volume per connection type (Bluetooth / 2.4GHz dongle). ANC has its own intensity memory on the device, separate from the existing Transparency level control, and was not previously adjustable at all.
• 4 new Sonar presets from SteelSeries GG 119.0.0: Bombanana!, Mortal Shell II, The Sinking City 2, Tukoni: Forest Keepers.

Source: GitHub