Audirvana Plus 2.5.1 with PCM->DSD Conversion

[mansr]

Would probably be helpful to see which filter options cause most CPU work with various computers.

 

There won't be much difference. Firstly, the resampler uses more CPU time than the SDM. Secondly, I haven't (yet) optimised the SDM for each filter.

[OldBigEars]

I'm intrigued to hear the new beta and have successfully downloaded it. But I can't figure out which settings I need to change. I'm on the "audio filters" tab and can see "To DSD" - but the button is greyed out and non-functional. Can someone help me out? Thanks!

[Pocobear]

I'm intrigued to hear the new beta and have successfully downloaded it. But I can't figure out which settings I need to change. I'm on the "audio filters" tab and can see "To DSD" - but the button is greyed out and non-functional. Can someone help me out? Thanks!

Under Preferences / Converter make sure that it says iZotope 64-bit SRC, then click on Advanced parameters, You should now be able to select To DSD.

[skatbelt]

I'm intrigued to hear the new beta and have successfully downloaded it. But I can't figure out which settings I need to change. I'm on the "audio filters" tab and can see "To DSD" - but the button is greyed out and non-functional. Can someone help me out? Thanks!

 

First, check your settings for native DSD Capability in the 'Audio System' tab. I had to choose DOP 1.1 for my Hugo there first.

[skatbelt]

Question: do the Advanced parameters for iZotope still matter when upsampling to DSD. If so, what happens exactly, do I filter or tweak the filter?

[OldBigEars]

First, check your settings for native DSD Capability in the 'Audio System' tab. I had to choose DOP 1.1 for my Hugo there first.

 

Thanks Skatbelt, that seemed to do the trick and it seems to work now. My DAC is a Concero HD - no issues so far and first impressions.....the sound is gorgeous. I'm curious about why DOP 1.1 vs DOP 1.0. Does anyone have any information about why you should choose one or the other?

[orgel]

Does the Mytek require drivers on Mac OS X?

 

Yes, the Mytek uses DICE drivers. Kind of a PITA. The recent replacement for the Stereo192-DSD, the Brooklyn, is driverless with OS X and Linux, so Mytek has learned the requisite lesson.

 

--David

[Jud]

Question: do the Advanced parameters for iZotope still matter when upsampling to DSD. If so, what happens exactly, do I filter or tweak the filter?

 

They still matter. Find the DSD filter you like best first (would be nice if you told us which one, what it is about the sound you like). Then you can see if any iZotope tweaks make things sound better.

[orgel]

Here is the new build with Mans new filters (you'll have 11 to choose from), they are not sorted by quality. But the Type A should be safer for DACs that have a tendency for noise.

https://audirvana.com/delivery/AudirvanaPlus_2.5.0.7.dmg

Please tell me if you encounter noise with a filter, the DAC model, and the song with which it happens.

 

I'm still experiencing the same symptoms with 2.5.0.7 as I described with 2.5.0.6. With 2.5.0.7, I tried Filter Types A0, A4, B0, B4, and C. None of these filters made a difference with the noise issues.

 

I did notice that when attempting to convert 24/192 PCM to DSD64, the spurious tones seemed to appear at exactly the same points in the track when trying different filters.

 

This time around, I also tried playing some DSF DSD64 tracks. Non-upsampled DSD64 plays without issue (do the different filters work in this case?). If I try to upsample DSD64 to DSD128, playback is fine, but there's no upsampling. This is shown by both A+ and the display on the Mytek.

 

When doing the PCM->DSD test, A+ would crash about every two filter changes, at the point when I tried to start playback after changing the settings.

 

My PCM test track is Track 1 ("Natsukashii") from Natsukashii by the Helge Lien Trio (24/192). The DSF test track is Track 1 ("These Are Soulful Days") from Lee-Way by Lee Morgan (2008 remaster, DSD64). DAC is Mytek Stereo192-DSD, latest firmware, latest FireWire drivers.

 

--David

[mansr]

I'm still experiencing the same symptoms with 2.5.0.7 as I described with 2.5.0.6. With 2.5.0.7, I tried Filter Types A0, A4, B0, B4, and C. None of these filters made a difference with the noise issues.

 

I did notice that when attempting to convert 24/192 PCM to DSD64, the spurious tones seemed to appear at exactly the same points in the track when trying different filters.

 

This time around, I also tried playing some DSF DSD64 tracks. Non-upsampled DSD64 plays without issue (do the different filters work in this case?). If I try to upsample DSD64 to DSD128, playback is fine, but there's no upsampling. This is shown by both A+ and the display on the Mytek.

 

When doing the PCM->DSD test, A+ would crash about every two filter changes, at the point when I tried to start playback after changing the settings.

 

My PCM test track is Track 1 ("Natsukashii") from Natsukashii by the Helge Lien Trio (24/192). The DSF test track is Track 1 ("These Are Soulful Days") from Lee-Way by Lee Morgan (2008 remaster, DSD64). DAC is Mytek Stereo192-DSD, latest firmware, latest FireWire drivers.

 

--David

 

Do you have another DAC you could try?

[orgel]

Do you have another DAC you could try?

 

Yes, but it's at the office (iDSD nano). It wouldn't surprise me at all if this turned out to be something peculiar to the Mytek. I'll give the nano a try on Monday.

 

--David

[jhwalker]

I've been playing with 2.5.0.7 this morning and greatly enjoying the sound, when it works

 

All tests performed at DSD128, have not tried DSD64 much, just to confirm it works.

 

2011 MacBook Pro i7, 2.5Ghz, 16GB RAM, 1TB SSD. LH Labs LightSpeed 2G USB cable, W4S Recovery, LH Labs Geek Pulse X Infinity (powered by LH Labs LPS), Alpha Prime.

 

Observations:

• DSD > DSD resampling does not work, just plays the original rate DSD, probably by design
  
• AU processing is not possible, AUs are simply ignored - is it possible to do the processing first to the PCM stream, then convert to DSD for output, so we can get the benefit of, for example, equalization and / or crossfeed?
  
• In general, the first 0.5 - 1 second of each track is cut off if you start playback on that track - not necessarily noticeable if there is a silence at the beginning, but consistent if the music starts at the beginning of the track (e.g., classical music intended to be gapless)
  
• Gapless works well; i.e., there is no cutoff of the first 0.5 -1 second of the track if proceeding from one track to another
  
• At all sample rates which are multiples of 44, there is a playback glitch / gap at 5 seconds into the playback. Very consistent with 44, 88, 176, 352 sample rates, and it always happens at 5 seconds, independent of the sample rate
  
• Sample rate that are multiples of 48 do not glitch at 5 seconds
  
• At least half of the time, playback hangs loading the first track. When this happens, CPU utilization is near 0; i.e., nothing is happening
  
• The playback hangs seem to be indiscriminate to the filter type, with a few caveats: A0, B0, and C seem to hang most frequently, but I've gotten them to play at least a few times - so there are no filters that simply don't work, ever. A2 and B1 seem to work every time.
  
• My "favorites" (when they work) are: B2, B3, B4, and C
  

 

Certainly looking forward to this capability when the hangs on playback and the cutoff of the first second and glitch at 5 seconds with the 44 multiples are corrected

[skatbelt]

They still matter. Find the DSD filter you like best first (would be nice if you told us which one, what it is about the sound you like). Then you can see if any iZotope tweaks make things sound better.

 

Thanks Jud. I will take some time for thorough listening and report back. For now I can tell that all filters work on a Chord Hugo without any problems (DSD64 as well as DSD128). The only thing is that each track starts with what would be best described as a fade in. So not full volume at once. Very short though and it does not appear between tracks when playing an album.

[jhwalker]

I've been playing with 2.5.0.7 this morning and greatly enjoying the sound, when it works

 

All tests performed at DSD128, have not tried DSD64 much, just to confirm it works.

 

2011 MacBook Pro i7, 2.5Ghz, 16GB RAM, 1TB SSD. LH Labs LightSpeed 2G USB cable, W4S Recovery, LH Labs Geek Pulse X Infinity (powered by LH Labs LPS), Alpha Prime.

 

Observations:

• DSD > DSD resampling does not work, just plays the original rate DSD, probably by design
  
• AU processing is not possible, AUs are simply ignored - is it possible to do the processing first to the PCM stream, then convert to DSD for output, so we can get the benefit of, for example, equalization and / or crossfeed?
  
• In general, the first 0.5 - 1 second of each track is cut off if you start playback on that track - not necessarily noticeable if there is a silence at the beginning, but consistent if the music starts at the beginning of the track (e.g., classical music intended to be gapless)
  
• Gapless works well; i.e., there is no cutoff of the first 0.5 -1 second of the track if proceeding from one track to another
  
• At all sample rates which are multiples of 44, there is a playback glitch / gap at 5 seconds into the playback. Very consistent with 44, 88, 176, 352 sample rates, and it always happens at 5 seconds, independent of the sample rate
  
• Sample rate that are multiples of 48 do not glitch at 5 seconds
  
• At least half of the time, playback hangs loading the first track. When this happens, CPU utilization is near 0; i.e., nothing is happening
  
• The playback hangs seem to be indiscriminate to the filter type, with a few caveats: A0, B0, and C seem to hang most frequently, but I've gotten them to play at least a few times - so there are no filters that simply don't work, ever. A2 and B1 seem to work every time.
  
• My "favorites" (when they work) are: B2, B3, B4, and C
  

 

Certainly looking forward to this capability when the hangs on playback and the cutoff of the first second and glitch at 5 seconds with the 44 multiples are corrected

 

Just a follow up:

 

Testing now with iFi Micro iDSD, direct from MacBook Pro (i.e., no reclocker or split cable, just direct from the MacBook to DAC), the glitch at 5 seconds is still consistent for all music at multiples of 44 sample rate. The failure to playback (i.e., no playback, stuck on "Loading first track . . . ") is, if anything, worse (or, let's say, "more consistent"): with the iFi, I have to pull out and reset the USB cable with each new playback session. That is, it will play back multiple tracks consecutively, but if I click "stop" to move on to another selection, playback will fail every time until I pull out / plug back in the USB cable, at which point, it will play fine until the next "stop".

 

So I'd think this has something to do with re-initializing the USB connection after "stop" - just speculating.

[AudioDoctor]

Just keeps getting better guys! Loving it!

[YashN]

Do we care about Foobar2000..?

 

Yes we do: For Windows users, it can play in bit-perfect mode, the GUI is highly configurable, it sounds great even on XP, it's free and all plugins to do DSD are free as well, including the configuration for it to do real-time PCM-to-DSD conversion.

[AudioDoctor]

Another overheating Mac Mini. Maybe they have inadequate cooling by design.

 

My Mini seems to not have this problem.

[AudioDoctor]

two SSDs 256 and 512gb

 

 

I have one SSD in mine, I think two would hold in more heat.

[AudioDoctor]

Mansr, .07 update cured the problem. By using all four cores now it spreads the heat load out. Now it is about the same as HQP, low to mid 70s with the fan running around 4500rpm. Thanks guys.Great work.

 

PS, using A0 and B0 so far.

 

Ah, I see.

[Jud]

orgel, what are your iZotope settings?

[skatbelt]

 

• AU processing is not possible, AUs are simply ignored - is it possible to do the processing first to the PCM stream, then convert to DSD for output, so we can get the benefit of, for example, equalization and / or crossfeed?
  

John, what makes you think that AU processing doesn't work?

 

Edit: Forget my question. I just experienced it myself. It makes this version not usable for me because I use an AU for parametric EQ to fight room modes. Comparing normal playback to upsampled DSD will be apples and oranges. I already wondered why every filter was so bloomy and bass heavy

[orgel]

orgel, what are your iZotope settings?

 

Steepness: 23

Filter max length: 1,900,000

Cutoff freq: 1.02

Anti-aliasing: 200.0

Pre-ringing: 0.55

 

I'm certainly open to making some changes if it addresses the PCM->DSD issue.

 

--David

[jhwalker]

Just a follow up:

 

Testing now with iFi Micro iDSD, direct from MacBook Pro (i.e., no reclocker or split cable, just direct from the MacBook to DAC), the glitch at 5 seconds is still consistent for all music at multiples of 44 sample rate. The failure to playback (i.e., no playback, stuck on "Loading first track . . . ") is, if anything, worse (or, let's say, "more consistent"): with the iFi, I have to pull out and reset the USB cable with each new playback session. That is, it will play back multiple tracks consecutively, but if I click "stop" to move on to another selection, playback will fail every time until I pull out / plug back in the USB cable, at which point, it will play fine until the next "stop".

 

So I'd think this has something to do with re-initializing the USB connection after "stop" - just speculating.

 

. . . and some additional feedback, now testing with the iFi Nano iDSD. Very similar behavior as with the Micro, as you'd expect, as they use the same firmware. Same gap at 5 seconds into each new playback session, same issue with initializing playback after a "stop", etc.

 

One more bit of data: both the Nano and Micro are capable of DSD256 playback with the latest firmware, and both sound great with it, except it appears my MBP is not capable of keeping up with the conversion, and Audirvana keeps running out of data in the buffer and pausing. Interesting point is that DSD256 uses almost exactly the same CPU as DSD128 (i.e., ~220-230% on my i7), so I wonder if there are some tweaks that could be made to the code to make DSD256 stable? i.e., it seems there is enough CPU capacity left over

[Outlaw]

DSD works ok.But it uses twice as much processor power then hqplayer.

[mansr]

DSD works ok.But it uses twice as much processor power then hqplayer.

That's because it's twice as good ;-)