INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1]   Go Down

Author Topic: 2CH-24bit-192kHz Problem with Switch Tracks: (anything other than Gapless)  (Read 3412 times)

d_pert

  • Regular Member
  • Galactic Citizen
  • ****
  • Posts: 398
  • I love music and great audio!

Hi,

I've just been playing around with some dynamite 2CH-24bit-192kHz WAV and APE made from a BluRay.

In 14.0.71 I'm getting intermittent blank-outs (like 1 second long) and premature endings (by like 1 second) only when Options > Track Changes > Switch tracks: (anything other than "Gapless") is hte setting.

I've put MC14 "through the mill" in terms of these settings and various bitrates/kHz source files and have never encountered this until I reached 192kHz. This includes AC3-WAV and DTS-WAV, 24/96 PCM 24/44.1 PCM, etc.

My "sound card" is a Steinberg MR816 pro interface with a 96kHx/24-bit operation max. I have MC's Output Format set to Stereo/24bit/96kHz/WASAPI.

No other problems.

Logged
Derek Pert
(Windows 11 Pro x64 / 32GB RAM)

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42442
  • Shoes gone again!

I can't think why the between track mode would matter.

Is it possible you're changing sample rate between files?  This forces a hard-restart of the playback chain so might cause a delay.

Otherwise, maybe try picking a longer buffering window.
Logged
Matt Ashland, JRiver Media Center

Alex B

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 10121
  • The Cosmic Bird

I think d_pert is correct and something is not working as it should.

I don't normally use 192 KHz files (such a resolution is just waste of disk space), but I created a small set of gapless 24/192 test samples. It contains five APE clips, about 10-15 seconds each. They play nicely gaplessly when MC's DSP is completely off and the playback mode is set to gapless, but when cross-fade is set the output is not correct. I tested cross-fade smooth at 0.1, 0.5 and 1 s. The symptoms are like d_pert described. MC's internal resampler at 96 KHz makes things worse and also the actual gapless mode has a problem with the first track that ends prematurely. Prebuffering was set to default (6 s).

The sample set is available here: http://rapidshare.com/files/288066301/192_24_gapless_samples.zip (31.3 MB)
Logged
The Cosmic Bird - a triple merger of galaxies: http://eso.org/public/news/eso0755

d_pert

  • Regular Member
  • Galactic Citizen
  • ****
  • Posts: 398
  • I love music and great audio!

Alex B: Thanks for the deeper probe into this. I have done more experiments, switching to/from 192/24 and 44.1/16 material, and trying every combination of settings under Playback Options > Audio, with and without and DSP plugins running, etc. Somehow, the problem has gone away ... all my settings are back to where they are by default ... in fact I use Norton Ghost to restore a flawless partition image with all my clean fresh install setups ... problem has not returned. The most annoying kind of intermittency!

Seems that just "messing around" made it go away ... probably temporarily.

I will report back when/if it reappears.

DP
Logged
Derek Pert
(Windows 11 Pro x64 / 32GB RAM)

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42442
  • Shoes gone again!

I haven't been able to reproduce this problem.

I'm using 192 KHz APE files, resampled to 96 KHz, output using WASAPI exclusive.  It's somewhat CPU intensive to do the resampling, but a Core Duo (or better) shouldn't have any problem with it.

Logged
Matt Ashland, JRiver Media Center

MusicHawk

  • Citizen of the Universe
  • *****
  • Posts: 796

Jumping in since this might be related to the problem I reported a few times with MC13 and MC14 -- what I describe as a 1-second audio drop-out when playing in an overlapping mode, just when the next track starts. The entire audio stream disappears briefly, then resumes with the new track playing while the old track finishes its ending. The problem might be happening all the time, but it's only audible when using a track overlapping mode.

(I use "cross-fade aggresssive" at "1 s" which thankfully does not actually cross-fade, it overlaps radio-station style allowing tracks to start/end naturally).

I think disabling DSP "fixes" the problem, which suggests that there's a processing delay in some circumstances that causes playback to stop briefly at the track transition point.

Almost all my files are MP3 (a few FLAC).

I found that the problem USUALLY happenes with 2-channel to 1-channel track transition (either way), but not consistently. Most of my 1-channel files also are lower bitrate than my 2-channel files, but I have plenty of the track combinations that flow just fine.

While fairly rare, the problem is reproducible between the same two files. During the same MC session I can replay the same tracks transition again and again and always get the same 1-second audio dropout.

There's a visible sympton: MC flashes "buffering" when the audio drops out. But what buffering? I diddled with the available buffering setting extensively with no audible change at all.

The problem happens on 3 computers of varying specs so I don't think it is a hardware problem.
Logged
Managing my media with JRiver since Media Jukebox 8 (maybe earlier), currently use Media Center for Audio/Music and Photos/Videos.
My career in media spans Radio, TV, Print, Photography, Music, Film, Online, Live, Advertising, as producer, director, writer, performer, editor, engineer, executive, owner. An exhausting but amazing ride.

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42442
  • Shoes gone again!

Jumping in since this might be related to the problem I reported a few times with MC13 and MC14 -- what I describe as a 1-second audio drop-out when playing in an overlapping mode, just when the next track starts. The entire audio stream disappears briefly, then resumes with the new track playing while the old track finishes its ending. The problem might be happening all the time, but it's only audible when using a track overlapping mode.

(I use "cross-fade aggresssive" at "1 s" which thankfully does not actually cross-fade, it overlaps radio-station style allowing tracks to start/end naturally).

I think disabling DSP "fixes" the problem, which suggests that there's a processing delay in some circumstances that causes playback to stop briefly at the track transition point.

Almost all my files are MP3 (a few FLAC).

I found that the problem USUALLY happenes with 2-channel to 1-channel track transition (either way), but not consistently. Most of my 1-channel files also are lower bitrate than my 2-channel files, but I have plenty of the track combinations that flow just fine.

While fairly rare, the problem is reproducible between the same two files. During the same MC session I can replay the same tracks transition again and again and always get the same 1-second audio dropout.

There's a visible sympton: MC flashes "buffering" when the audio drops out. But what buffering? I diddled with the available buffering setting extensively with no audible change at all.

The problem happens on 3 computers of varying specs so I don't think it is a hardware problem.


If the output format changes between tracks, you won't be able to cross-fade.

You can avoid format changes by using the 'Output Format' DSP to mix everything to the same format.
Logged
Matt Ashland, JRiver Media Center

Alex B

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 10121
  • The Cosmic Bird

If the output format changes between tracks, you won't be able to cross-fade.

You can avoid format changes by using the 'Output Format' DSP to mix everything to the same format.

I was about to post the same answer, but you were quicker.

I haven't been able to reproduce this problem.

I'm using 192 KHz APE files, resampled to 96 KHz, output using WASAPI exclusive.  It's somewhat CPU intensive to do the resampling, but a Core Duo (or better) shouldn't have any problem with it.

Did you try Direct Sound on XP?

The problem I was able to reproduce was with using my 24/192 gapless samples (no format change from one track to another) and without MC's resampler or any other DSP, except the cross-fader. I used the Direct Sound output mode so that XP's Kernel Mixer "helpfully" and automatically resampled the final output to 48 KHz (despite the fact my sound card can do 96 KHz). XP's built-in resampler is designed to be light and it doesn't use much CPU so I don't think it was causing the problem.
Logged
The Cosmic Bird - a triple merger of galaxies: http://eso.org/public/news/eso0755

MusicHawk

  • Citizen of the Universe
  • *****
  • Posts: 796

Matt, thanks for the tip. I was already using Output Format, and I think it helped reduce the number of track transition problems, but not eliminate them.

I've long suspected that the order of DSP stages matters (as it does in analog audio), and worried that Output Format was possibly happening in the order indicated in the DSP list, putting it in the middle of the audio flow (and equally bad, putting Volume Levelling at the end)

I just used MC's new ability to re-order to set Output Format as stage 1, then Volume Leveling as stage 2, then whatever else I might want to use. With fingers crossed, this might be the solution...

Logged
Managing my media with JRiver since Media Jukebox 8 (maybe earlier), currently use Media Center for Audio/Music and Photos/Videos.
My career in media spans Radio, TV, Print, Photography, Music, Film, Online, Live, Advertising, as producer, director, writer, performer, editor, engineer, executive, owner. An exhausting but amazing ride.

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42442
  • Shoes gone again!

Matt, thanks for the tip. I was already using Output Format, and I think it helped reduce the number of track transition problems, but not eliminate them.

I've long suspected that the order of DSP stages matters (as it does in analog audio), and worried that Output Format was possibly happening in the order indicated in the DSP list, putting it in the middle of the audio flow (and equally bad, putting Volume Levelling at the end)

I just used MC's new ability to re-order to set Output Format as stage 1, then Volume Leveling as stage 2, then whatever else I might want to use. With fingers crossed, this might be the solution...



Output Format is special.  It always happens first, and changes require a restart of playback to take effect.

Replay Gain happens as data is added to the input buffer, so changes are delayed six or so seconds.

Dolby Digital encoding happens last as data is delivered to the output.

All other DSPs happen just-in-time as data is delivered to the output plugin.  The user configured order is respected.
Logged
Matt Ashland, JRiver Media Center

Alex B

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 10121
  • The Cosmic Bird
Re: 2CH-24bit-192kHz Problem with Switch Tracks: (anything other than Gapless)
« Reply #10 on: October 05, 2009, 02:48:26 pm »

Output Format is special.  It always happens first, and changes require a restart of playback to take effect.

This is good and necessary for the channel and sample rate settings, but isn't the bit depth drop from the internaly used 32-bit format to the 24 or 16 bit integer format always the last step in the DSP chain? (Probably it is, and you just wanted to simplify your reply.)

Dolby Digital encoding must naturally be the final step if it is enabled because the signal cannot be modified after it, but is more like a postprocessor than part of the standard DSP chain.
Logged
The Cosmic Bird - a triple merger of galaxies: http://eso.org/public/news/eso0755

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42442
  • Shoes gone again!
Re: 2CH-24bit-192kHz Problem with Switch Tracks: (anything other than Gapless)
« Reply #11 on: October 05, 2009, 06:02:14 pm »

This is good and necessary for the channel and sample rate settings, but isn't the bit depth drop from the internaly used 32-bit format to the 24 or 16 bit integer format always the last step in the DSP chain? (Probably it is, and you just wanted to simplify your reply.)

You're right (of course).

I'm rather certain if Alex and I both wrote a white-paper about Media Center, his would be more accurate and informative.  But mine would have more pictures of farm animals.
Logged
Matt Ashland, JRiver Media Center
Pages: [1]   Go Up