Release Candidate 4.6 Hotfix — Clearer Audio, Reliable Feedback & Disconnects

Release Candidate 4.6 Hotfix — Clearer Audio, Reliable Feedback & Disconnects

RC4.6 brings together the fixes since RC4.5.9: more reliable controller audio, safer feedback delivery under load, and cleaner Bluetooth disconnects. The faster DualSense feedback path from RC4.5.9 stays in place.

Controller audio that starts and stays on the right source

• DualSense speaker recovery: restarting or switching audio sources no longer stalls speaker warmup because an earlier session's counters were reused.
• Sonar compatibility: physical DualSense USB and Bluetooth capture use the established Sonar-compatible endpoint capture policy, including audio from an emulated DualShock 4 headset.
• Keep the source quality: app and external-endpoint capture for DualSense uses 48 kHz floating-point audio. The virtual DualShock 4 headset endpoint remains at its native 32 kHz; physical DualShock 4 Bluetooth audio keeps its existing codec and rate.
• Correct USB playback: different source rates are converted to the physical DualSense output rate instead of being copied frame-for-frame. Large callbacks are no longer truncated, and small callbacks are processed without waiting to fill an extra block.
• Safer source changes: old callbacks and buffered samples cannot leak into a replacement source. Capture shutdown and shared-app capture retirement no longer hold the manager lock while waiting on callbacks.

Feedback delivery beyond DualSense

• Physical DualShock 4: fresh motor changes and explicit stops can pass immediately during Bluetooth audio instead of waiting behind the unchanged-state maintenance interval.
• Physical DualShock 3: a failed output write remains pending for retry; a newer stop can replace the failed active effect.
• Switch 2 Pro and Joy-Con 2: eligible identical commands at an unclaimed pending queue tail can share an entry. Changed effects and stops, A-to-B-to-A sequences, PCM and timed feedback are not folded together. This is a delivery change, not a rumble gain, encoding or cadence retune.
• Virtual DualShock 4 / VIIPER: the V3 speaker-output path reserves room for a motor-stop command, preserves ordered changes, and reports a full distinct-command queue instead of silently accepting and dropping its contents. Reconnecting feedback readers cannot let an old connection retire the new owner's state.

Disconnect and profile safeguards

• No stale controller row after manual Bluetooth disconnect: the controller-list and tray commands now request complete removal, rather than depending on a later HID read failure after transport cleanup has already finished.
• Reject stale Game Bar activation: delayed compatibility activation rechecks the current profile policy before allocating or rerouting output. Switching to Xbox output, disabling compatibility or selecting input-only mode prevents an old activation request from taking over.

Packages and updating

• DS4Windows VIIPERRC4.6, Windows file/installer version 5.0.6.0.
• Bundled VIIPER 0.1.5-rc4.6, with matching source and notices. USB/IP remains 0.9.7.7.
• Complete offline installer and portable ZIP, including VIIPER and the required Xbox output identity. Portable users should extract the entire folder.
• The published DS4Updater 2.0.6 remains compatible; no replacement updater is needed for this version.

Validation and limits

This is an unsigned release candidate. The rebuilt local x64 suite passed 5,717 tests with zero failures, including eight isolated DS4Windows-to-VIIPER process tests; three live-audio tests remain explicitly gated. Regression coverage exercises speaker priming, sample-rate continuity and channel placement, source retirement, feedback order and stop admission, reconnect ownership, manual removal, and Game Bar activation policy. Allocation-sensitive assertions remain enabled; a traced GC-counter measurement issue was corrected in the test harness without raising the zero-byte limit or changing production behavior.

Software tests are not a promise of unlimited controller output rates or zero physical latency. Final audio listening, disconnect behavior and gameplay feel still need tester confirmation. The rare once-in-a-day movement interruption remains unattributed; the Game Bar safeguard is not being presented as proof that incident is solved. Genuine disconnect and shutdown neutral-report safety remains intact.

Detailed source evidence: cross-controller feedback, audio and Sonar, manual disconnect, and rare input investigation.