Dear Maintainer, I can independently confirm this regression on Debian 13 (Trixie) using a MOTU M2 audio interface.
Hardware: Motherboard: MSI MAG B660 Tomahawk WiFi DDR4 CPU: Intel Core i7-12700 USB controller: Intel xHCI Audio interface: MOTU M2 [07fd:000b] USB audio driver: snd_usb_audio Kernel comparison: 6.12.113+deb13-amd64: WORKING 6.12.115+deb13.1-amd64: BROKEN The failure initially appeared following an upgrade from Debian 12 Bookworm, kernel version 6.1.54 to Debian 13 Trixie. After extensive troubleshooting, I tried installing and booting Linux 6.12.113. Audio playback through the MOTU M2 was restored, without changing the Debian 13 userspace or audio configuration. Symptoms under 6.12.115: -ALSA successfully recognizes the MOTU M2, and PipeWire exposes the playback device. -When the MOTU is selected as the default audio output, there is no sound from the monitor headphones or speakers. -When selected as the default audio output, attempting to play videos in Firefox such as YouTube results in playback stalling indefinitely. -Selecting my Bluetooth headphones or HDMI audio outputs restored normal YouTube playback. -Switching the MOTU to PipeWire's Pro Audio profile allowed YouTube playback to proceed, but no audio is produced. -Direct ALSA playback remains silent even after stopping PipeWire and WirePlumber. Direct ALSA reproduction: After stopping WirePlumber and PipeWire: speaker-test -D plughw:3,0 -c 2 -r 48000 -t sine -f 440 The playback stream reports RUNNING, but the ALSA hardware pointer never advances. Output of /proc/asound/card3/pcm0p/sub0/status: state: RUNNING delay: 48000 avail: 0 avail_max: 0 hw_ptr: 0 appl_ptr: 48000 The same values were observed after waiting several seconds. Additional troubleshooting notes: -Reloading snd_usb_audio with lowlatency=0 did not resolve the problem. -Running ALSA capture simultaneously with playback using arecord and speaker-test did not cause hw_ptr to advance -The MOTU's USB audio playback endpoint uses asynchronous transfers and implicit feedback. -ALSA arecord recordings produce nonzero audio samples. A controlled input test showed an approximately 15.2 dB RMS increase on channel 2 (mic channel) when producing an input signal. Although playing the .wav file back through my headphones it sounded silent. -The input meters on the MOTU M2 audio interface itself respond normally. -No obvious USB audio errors were observed in the kernel logs during the playback failure. -The M2 and M4 share USB vendor/product ID 07fd:000b, and this appears to be the same regression described in this report. Thank you. -Christopher Ludwin

