Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[mikeza]

Hi all,

I'm getting weird audio glitches from JRiver that sound like a very brief  (~1s?) volume burst. This occurs about every 2-10(?) minutes. I thought it was something else in my signal chain, but switching from JRiver to mpv solved the glitch.

My signal chain looks like JRiver -> ALSA (via Pulse) -> Scarlett 2i4 (via USB) -> Powered Studio Monitors (Balanced lines)

I'm not sure I've ever seen this issue from JRiver when not outputting to the Scarlett 2i4 (i.e. I don't think I hear it when playing from my laptop speakers but the audio qualities not great so I'd have to double check, I can also try headphones from the aux jack but I don't think it will occur). I have used JRiver with the Scarlett in the past without any issues.

Troubleshooting/reproduction:
- Running a CPU stressor does not recreate the glitch

Attempted solutions:
- Full system update
- JRiver update
- Disabled all options in the DSP studio
- Moved prebuffering from 6 -> 10 -> 20s
- Increased playback latency to 250ms

System info

Code: [Select]

mz@mzxps:~/ > archey           

               +               
               #               
              ###               
             #####             
             ######             
            ; #####;            User: mz
           +##.#####            Hostname: mzxps
          +##########           Distro: Arch Linux
         #############;         Kernel: 4.19.8-arch1-1-ARCH
        ###############+        Uptime: 1 day, 11:08
       #######   #######        Window Manager: i3
     .######;     ;###;`".      Desktop Environment:
    .#######;     ;#####.       Shell: /bin/zsh
    #########.   .########`     Terminal: rxvt-unicode-256color
   ######'           '######    Packages: 1406
  ;####                 ####;   CPU: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz
  ##'                     '##   RAM: 3490 MB / 7857 MB
 #'                         `#  Disk: 776GB / 1019GB


Code: [Select]

Library
    Total files: 36508
    Audio files: 36419
    Image files: 0
    Video files: 21
    Other files: 68
Processing
    Thumbnails built: 100% (36504 of 36508)
    Audio analyzed: 100% (36419 of 36440)
Background Tools Running
    No tools currently running
Power
    No entries that affect the system power state
Media Center
    Version:
    Install path: /usr/lib/jriver/Media Center 24/
    Interface plug-ins:
    JRMark: never run
    Memory used:
    Handles used:
Advanced System Info
    Memory & CPU
        CPU features: MMX, SSE2, SSE3, SSSE3, SSE4.1
        Memory Available:  3.3 GB
        Memory Size:       7.5 GB
        Total Logical Cores: 4
    Operating System
        Linux - 64 bit
Benchmark

Code: [Select]

=== Running Benchmarks (please do not interrupt) ====== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
    Single-threaded integer math... 4.939 seconds
    Single-threaded floating point math... 3.411 seconds
    Multi-threaded integer math... 2.718 seconds
    Multi-threaded mixed math... 1.796 seconds
Score: 1477

Running 'Image' benchmark...
    Image creation / destruction... 0.423 seconds
    Flood filling... 0.319 seconds
    Direct copying... 0.649 seconds
    Small renders... 1.131 seconds
    Bilinear rendering... 2.301 seconds
    Bicubic rendering... 1.378 seconds
Score: 3548

Running 'Database' benchmark...
    Create database... 0.311 seconds
    Populate database... 1.007 seconds
    Save database... 0.150 seconds
    Reload database... 0.093 seconds
    Search database... 1.019 seconds
    Sort database... 0.799 seconds
    Group database... 0.550 seconds
Score: 5471

JRMark (version 24.0.68 64 bit): 3499
Here's a log during the glitch

https://www94.zippyshare.com/v/petpkjBq/file.html

[Awesome Donkey]

Any particular reason why you're using Pulse? Reason I ask is because it resamples everything to 48 kHz. I'd recommend the hw: or front: outputs if you have them available.

[mikeza]

Any particular reason why you're using Pulse? Reason I ask is because it resamples everything to 48 kHz. I'd recommend the hw: or front: outputs if you have them available.

In general, pulse is just more convenient when switching audio devices frequently (which I do since it's a laptop). However, I get a playback error if I try to select the Scarlett via ALSA directly. Not sure why.

I've set to 2 chan and 4 chan outputs and neither works

Code: [Select]

mz@mzxps:~/ > cat /proc/asound/card1/pcm0p/sub0/hw_params
access: MMAP_INTERLEAVED
format: S32_LE
subformat: STD
channels: 4
rate: 44100 (44100/1)
period_size: 32768
buffer_size: 65536


[bob]

Try the latest build and enable thread priorities for your user or group. See the announcement for 24.0.69

[mikeza]

Try the latest build and enable thread priorities for your user or group. See the announcement for 24.0.69

I've added myself to the realtime group (and installed realtime-privileges) but I'm still getting skips on the newest build. It appears that the mediacenter24 threads are running at a nice value of 10 which is low priority?

Edit: If I manually renice a couple mediacenter24 processes to -10, the glitches still occur.

Just a note that these are far more of a volume boost than a 'skip'.

Edit2: I think it may have something to do with the scarlett having 4 channels

[bob]

I've added myself to the realtime group (and installed realtime-privileges) but I'm still getting skips on the newest build. It appears that the mediacenter24 threads are running at a nice value of 10 which is low priority?

Edit: If I manually renice a couple mediacenter24 processes to -10, the glitches still occur.

Just a note that these are far more of a volume boost than a 'skip'.

Edit2: I think it may have something to do with the scarlett having 4 channels

During playback you should see one thread marked RR

[mikeza]

Changing to a 2 channel output within a 4 channel container both allows me to play directly to the card via ALSA and removes skips. I'm not sure why a straight 4 channel output doesn't have the same effect.