...

Any idea on how this frame number can be read,

With usb_get_current_frame_number().

I hacked something to refine the delay information based on the USB frame number. see attached patch. Two points can be improved: - I am not sure how to convert from USB frames to ALSA frames. I used a fixed value based what my headset supports - Every second or so the estimate seems to be off by 96 samples, see below. Not sure why this happens, how accurate the counter is and if there a delay in handling retired urbs? Looks good enough though. Thanks, -Pierre

[10880.896486] plb-urb -- actual delay 96 estimated delay 0 [10881.917729] plb-urb -- actual delay 48 estimated delay 0 [10882.937948] plb-urb -- actual delay 96 estimated delay 0 [10883.966190] plb-urb -- actual delay 96 estimated delay 0 [10884.987425] plb-urb -- actual delay 96 estimated delay 0 [10886.008654] plb-urb -- actual delay 96 estimated delay 0 [10887.029931] plb-urb -- actual delay 96 estimated delay 0 [10888.050153] plb-urb -- actual delay 96 estimated delay 0 [10889.071375] plb-urb -- actual delay 96 estimated delay 0