HQ Player

[AnotherSpin]

When I wrote yesterday my post http://www.computeraudiophile.com/f11-software/hq-player-20293/index79.html#post423292 I forgot to mention I had [x] DirectSDM checked on the DSDIFF/DSF Settings dialog.

 

I am playing 44.1k/16b and 96k/24b recordings. SDM(DSD) output is chosen, my DAC shows DSD128 (through DoP).

 

Please trace the differences a) and b) in processing with such PCM to DSD playback when:

 

1. dither setting is for example NS5

a) [x] DirectSDM is checked

b) [ ] DirectSDM is not checked (volume is set to -3dB)

 

2. [x] DirectSDM is checked

a) NS5 is set

b) NS9 is set

 

3. [x] DirectSDM is checked

a) none 'Filter' set in 'PCM defaults'

b) poly-sinc set as 'Filter' in 'PCM defaults'

 

My result is: In the case of DirectSDM checked both PCM defaults and SDM defaults are in action. My guess is that with [x] DirectSDM PCM type of upsampling is done to the rate required for DSD playback rate and then PCM is converted to DSD without DSD type of upsampling.

Why do I think that:

You say with PCM to DSD 'PCM defaults' 'Filter' has no effect.

But: With [x] DirectSDM checked

1. I am setting 'Filter' to none and playback starts immediately (less than 1 second).

2. I am setting 'Filter' to poly-sinc and it tooks about 15 seconds on my notebook for playback to start, it is clear that the PCM filter was initialized! During that period, processor utilisation went to 100%, as usual. Please note I did not change the filter in the main window, but in the Settings dialog.

 

Then I can repeat the sequence 1. and 2. with the same result ...

 

I have the same or almost the same in my case with Mac OS/Loki

[AnotherSpin]

Miska, now I have the following settings for Loki, DirectSDM checked. Is it correct?

 

[ted_b]

Bogi, I am under the impression that checking DirectSDM is irrelevant to any PCM sources, whether upsampling in PCM or converting to SDM/DSD. DirectSDM is ONLY a setting for DSD-based material/sources. When DirectSDM is checked, DSD-based sources are left alone and played back without any processing whatsoever. Period. The checkbox is irrelevant for PCM anything. Correct? (I could easily be wrong )

[AnotherSpin]

Bogi, I am under the impression that checking DirectSDM is irrelevant to any PCM sources, whether upsampling in PCM or converting to SDM/DSD.

 

Sorry for answering instead of Bogi, DirectSDM checking/un-checking makes impossible/possible to set PCM defaults filter to "none".

[bogi]

Ted, I and AnotherSpin disagree with Miska on description of HQPlayer behaviour when [x] DirectSDM is checked and PCM to DSD conversion is done.

 

[x] DirectSDM sounds me very well with SACD rips and I often forget that I have this set. I start to play some PCM recording with [x] DirectSDM still set and then I am observing things Miska clearly didn't count with. My last picture quite on the end of previous thread page clearly shows a bug - SDM(DSD) output is chosen, but selection of PCM bitrates is provided and 44.1k is displayed and played on my DAC.

 

It is not convenient to change settings everytime when I change from DSD to PCM recording and back. I like the DirectSDM choice for SACD rips. I would like from HQPlayer to behave this way: If DirectSDM is set, but PCM is played and SDM(DSD) output is chosen, the DirectSDM setting should be completely ignored by HQPlayer. The same behaviour as without DirectSDM should be applied. Today that's not true and that's the base of my previous posts.

 

If [x] DirectSDM is checked and PCM to DSD is performed in HQPlayer, PCM filters are in use. I clearly demonstrated that with examples of different filter initialization times as well as with the picture which shows PCM bitrates while SDM(DSD) output is chosen.

 

So HQPlayer contains a bug - for that special case [x] DirectSDM and PCM to DSD.

[bogi]

Sorry for answering instead of Bogi, DirectSDM checking/un-checking makes impossible/possible to set PCM defaults filter to "none".

 

I wrote in a previous post (the last on the previous thread page) that when I have set 'PCM defaults' 'Filter' to none and when I have [x] Direct SDM set, 44.1k (for CD rip) is output to my DAC even if I have set SDM(DSD) in the main window. That' the bug I reported. Your Loki as DSD only DAC is unable to play PCM, so you cannot get any sound in this case.

[AnotherSpin]

I wrote in a previous post (the last on the previous thread page) that when I have set 'PCM defaults' 'Filter' to none and when I have [x] Direct SDM set, 44.1k (for CD rip) is output to my DAC even if I have set SDM(DSD) in the main window. That' the bug I reported. Your Loki as DSD only DAC is unable to play PCM, so you cannot get any sound in this case.

ok)

[ted_b]

Bogi, now I understand. Thanks. I will have to check that myself on my copy. Sorry for the further confusion.

[tranz]

If you do any processing, like upsampling or volume control, you should have some dither selected. This is because the calculated sample values will have full 64-bit floating point precision and DAC cannot utilize such precision, it needs to be limited to 24- or 32-bit values. Doing this value range limiting without dither produces quantization distortion.

 

So the only case when dither is not needed is bit-perfect output.

 

Hi Miska,

 

Thank you, I forgot about the digital volume impact.

 

Because digital volume control is in play, which drops bits, dithering is needed. And perhaps, because all streams go through the limiter logic, they are no longer bit-perfect even with everything set to "none" and 0dB. Might be something to check out.

 

Cheers

[Miska]

Because digital volume control is in play, which drops bits, dithering is needed. And perhaps, because all streams go through the limiter logic, they are no longer bit-perfect even with everything set to "none" and 0dB. Might be something to check out.

 

With upsampling (or downsampling), the calculated sample values will have more precision than any current DAC has, so some bits need to be dropped. That's why dither is needed.

[Miska]

Ted, I and AnotherSpin disagree with Miska on description of HQPlayer behaviour when [x] DirectSDM is checked and PCM to DSD conversion is done.

 

[x] DirectSDM sounds me very well with SACD rips and I often forget that I have this set. I start to play some PCM recording with [x] DirectSDM still set and then I am observing things Miska clearly didn't count with. My last picture quite on the end of previous thread page clearly shows a bug - SDM(DSD) output is chosen, but selection of PCM bitrates is provided and 44.1k is displayed and played on my DAC.

 

It is not convenient to change settings everytime when I change from DSD to PCM recording and back. I like the DirectSDM choice for SACD rips. I would like from HQPlayer to behave this way: If DirectSDM is set, but PCM is played and SDM(DSD) output is chosen, the DirectSDM setting should be completely ignored by HQPlayer. The same behaviour as without DirectSDM should be applied. Today that's not true and that's the base of my previous posts.

 

If [x] DirectSDM is checked and PCM to DSD is performed in HQPlayer, PCM filters are in use. I clearly demonstrated that with examples of different filter initialization times as well as with the picture which shows PCM bitrates while SDM(DSD) output is chosen.

 

So HQPlayer contains a bug - for that special case [x] DirectSDM and PCM to DSD.

 

I have to say that I don't understand what is the exact problem...

 

1) Filter=none, DirectSDM=enabled -> output format follows source format

2) Filter=something, DirectSDM=enabled -> output format follows what ever selected in main window, but DSD it output bit-perfect in SDM output mode

3) Filter=none, DirectSDM=disabled -> this is a very messy case, in PCM output format things follow source PCM format, in SDM output format case DSD sources are also rate converted to requested output rate

4) Filter=something, DirectSDM=disabled -> output is either PCM or DSD constantly to the output format requested

 

My personal normal use case is to have DirectSDM enabled and output in SDM mode. This results in PCM sources output at requested DSD rate and DSD sources output bit-perfect at source rate. No surprises in this case.

 

 

P.S. It is very tempting to remove the "none" PCM filter altogether. Originally it kept output in PCM mode always, but then people requested that DSD should still be played as DSD which made all the hell break loose in terms of complexity.

[Miska]

Miska, now I have the following settings for Loki, DirectSDM checked. Is it correct?

 

Looks OK, the "DAC Bits" doesn't have effect here, but it doesn't harm either. Same goes for Bitrate setting which is limited to 3072000 by the DAC. What makes sense is to experiment which one sounds better to you, 2822400 or 3072000 output rate...

[Miska]

Miska, I would yet to add the following to my previous post, please check it self:

 

What you have in the second screenshot shouldn't be possible at all.

 

2. [x] DirectSDM is checked. PCM bitrates are unexpectedly showing. DAC shows 44.1k. That indicates me bug and it corresponds to what I wrote previously.

 

This certainly shouldn't happen. I would like to know how exactly you manage to end up in this situation. Can you export and email me your settings file if there are no special steps needed to reproduce this?

[AnotherSpin]

Looks OK, the "DAC Bits" doesn't have effect here, but it doesn't harm either. Same goes for Bitrate setting which is limited to 3072000 by the DAC. What makes sense is to experiment which one sounds better to you, 2822400 or 3072000 output rate...

 

Thank you!

[bogi]

I have to say that I don't understand what is the exact problem...

 

I’m very sorry but you did not read my posts enough carefully. I pointed you to the case when source is PCM and output is set to SDM(DSD). Please look to this case in more detail.

 

1) Filter=none, DirectSDM=enabled -> output format follows source format

 

When source is PCM and output is set to SDM(DSD), HQPlayer is outputting source rate PCM for example 44.1k. Further it provides PCM sample rates for selection, although SDM(DSD) output is selected (my previous picture). That’s quite against logic of HQPlayer settings and is quite confusing.

 

2) Filter=something, DirectSDM=enabled -> output format follows what ever selected in main window, but DSD it output bit-perfect in SDM output mode

 

Yesterday you wrote:

Yes, that's the case, the two boxes are mutually exclusive. "PCM Defaults" is used when output format is PCM, "SDM Defaults" is used when output format is SDM. Regardless of input (source) format.

 

Again, please look at the case when source is PCM and output is set to SDM(DSD). You wrote that only settings from the SDM defaults box have to apply. But in this specific case both I and AnotherSpin are pointing to you our listening experience and today I added also additional technical information about delays when filters are initialized. These delays clearly show that settings from the PCM defaults box are applied. You didn’t read my previous posts carefully and you didn’t understand that, although I and AnotherSpin are pointing you here already the 2nd day.

 

I am repeating at this place what I already wrote few hours ago:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

You say with PCM to DSD 'PCM defaults' 'Filter' has no effect.

But: With [x] DirectSDM checked

1. I am setting 'Filter' to none and playback starts immediately (less than 1 second).

2. I am setting 'Filter' to poly-sinc and it tooks about 15 seconds on my notebook for playback to start, it is clear that the PCM filter was initialized! During that period, processor utilisation went to 100%, as usual. Please note I did not change the filter in the main window, but in the Settings dialog.

 

Then I can repeat the sequence 1. and 2. with the same result ...

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

Please note I am pointing to setting filter type in the PCM defaults box and not in SDM default box. Did you start to understand please? The filter initialization delays could not be such if PCM defaults are not applied!

 

The fact that PCM defaults apply in [x] Direct SDM, PCM input and SDM(DSD) output mode, is yet more explained in this post http://www.computeraudiophile.com/f11-software/hq-player-20293/index80.html#post423742 Please read it carefully.

 

Because of PCM defaults are applied in this case, you are wrong also in the following:

 

Dither code is not called at all when output is SDM because it is not relevant for that case at all... Same goes for the DAC bits setting.

 

This is not true in this specific case. We found that PCM defaults are applied including dither and DAC bits.

 

3) Filter=none, DirectSDM=disabled -> this is a very messy case, in PCM output format things follow source PCM format, in SDM output format case DSD sources are also rate converted to requested output rate

 

In the case of PCM to DSD this setting causes no troubles. When DirectSDM=disabled, PCM defaults have really no effect.

 

4) Filter=something, DirectSDM=disabled -> output is either PCM or DSD constantly to the output format requested

 

This setting is without any issue and here I agree with your description how HQPlayer works.

 

My personal normal use case is to have DirectSDM enabled and output in SDM mode. This results in PCM sources output at requested DSD rate and DSD sources output bit-perfect at source rate. No surprises in this case.

 

That’s what I want. The only problem is that you are saying PCM defaults are not used in this case and I with AnotherSpin are pointing you to information, that PCM defaults are in action. If it is still not clear, please go back to my explanation of delays in filters initialization. Look at my picture with that PCM sample rate bug and think about how it could occur. Make a HQPlayer run with trace and you will see PCM defaults are in action, when DirectSDM=enabled, source is PCM and output mode is SDM(DSD).

[tranz]

Well Miska, I have to say, amazing job on the filters. You wrote all the code for those?

 

Just using poly-sinc, NS9, 24 bits setting, 192Khz PCM and playing back a ripped CD spanked Amarra which until now was my go to baseline for RBCD.

 

I have not heard better, period!

...

 

Well, after having lived with HQP a bit longer I may have jumped the gun with my post. Earlier this year I had purchased a license but did not have the time to test until this week. My initial reaction was using an old CD and the level of detail really was incredible. But then for longer listening and other material it became too hot/bright especially apparent on female vocalists and am back to Amarra.

 

And although not the point of HQP it does worry me a bit that bit-perfect playback was not possible on the Mac.

 

However, I will have to try again when I find a way to turn off all the internal DAC filtering and am able to use the high upsampling method to PCM 384 and DSD 256 as it seems that is where the magic happens.

 

Miska's support here is amazing and everyone's input to troubleshooting are much appreciated.

 

Cheers

[ted_b]

So, I have a question about conventional wisdom on upsampling:

Now that I have a tool like HQplayer that allows incredible flexibility to get upsampling right, I wondered about what HQplayer users (esp Jussi) use as telltale signs that a certain sample rate is the target for upsampling. Is it always the choice to first simply go to the max rate of the DAC, and if DSD-capable simply the max DSD bit rate? What about the idea that max is usually never a good goal in anything audio...wouldn't max (as a standard) always involve the risk of "outside of spec" kinds of problems, pushing the limits? Is max always the sweetspot or do folks listen for something special (pace, timbre, noise) to assign targets?

[clipper]

I have few questions about HQ Player:

 

 

2. Can it be used with AO/JPlay on a CAPS machine.

 

 

 

I have another CAPS setup with AO/JPlay with USB going to DAC.

 

Thanks.

 

This (HQ Player with AO and JPLAY) does work and it sounds great.

 

I usually use a dual pc setup, with HQ Player running on a Windows 8.1 machine with JPLAY (control pc) sending to a 2012 Server (core mode)/AO/JPLAY (hibernate mode on) machine (audio pc).

 

But I just tested by installing HP Player on the 2012 Server/AO/JPLAY machine, and it works well.

 

I'm using the 1.31 beta of AO, which I'm pretty sure (from reading) is necessary for this to work "out of the box."

 

JPLAY version is JPLAY 6.

[AnotherSpin]

The only problem is that you are saying PCM defaults are not used in this case and I with AnotherSpin are pointing you to information, that PCM defaults are in action. If it is still not clear, please go back to my explanation of delays in filters initialization. Look at my picture with that PCM sample rate bug and think about how it could occur. Make a HQPlayer run with trace and you will see PCM defaults are in action, when DirectSDM=enabled, source is PCM and output mode is SDM(DSD).

I have no doubts PCM defaults are effected during PCM up-sample to DSD and I found out through experience that left window settings should be used to change sound characteristics. In my setting, which includes Mac OS and Loki DSD DAC dithers NS4, NS5, NS9 all sounds differently, even more, I found out that I prefer dither set to "none". The same with PCM defaults filters, they give different sound characteristics.

[bogi]

I made yet one experiment with PCM recording and SDM(DSD) output setting.

[x] DirectSDM was enabled and minringFIR Filter set in both PCM defaults and SDM defaults. I set this filter because it causes low processor load.

Now when I exchanged filters in the main window, for example with poly-sinc, it took those 15 sec to initialize. When I changed back to low load filter like the 2s filters or minringFIR, playback started within 1 second. So it seems the SDM filters from the main window were really initialized.

 

I could wrongly interpret delays in filter initialization in my previous posts. It is not so easy to find out what is happening and it costs time.

 

After this result I am unsure if those influences of 32/24 DAC bits and dither settings were not related only to the case [x] DirectSDM was enabled and PCM defaults 'Filter'=none. In that case unexpectedly PCM was output to my DAC, although SDM(DSD) output was set in the main window. I discovered this unexpected fact too late and I am now unsure which of my listening results are related exactly to this situation.

 

I will let this topic open for a week and make yet some experiments and listening tests. The case [x] DirectSDM=enabled, Filter=none, PCM source and SDM(DSD) output should be avoided from tests because it causes misleading behavior, so results can be then misinterpreted.

[skerry2006aj]

Hello!

 

I am working at a project that does a similar thing like HQplayer, but with a different algorithm for resampling, it contains an algorithm for transient recovery, and a very different, intuitive and elegant user interface!

 

I invite you to my project's page!

 

http://www.computeraudiophile.com/f11-software/i-am-building-ideal-free-music-playback-software-24373/#post423098

[ted_b]

Hello!

 

I am working at a project that does a similar thing like HQplayer, but with a different algorithm for resampling, it contains an algorithm for transient recovery, and a very different, intuitive and elegant user interface!

 

I invite you to my project's page!

 

http://www.computeraudiophile.com/f11-software/i-am-building-ideal-free-music-playback-software-24373/#post423098

 

Seriously? Let's keep discussing this thread's subject.

[AnotherSpin]

 

I am working at a project that does a similar thing like HQplayer, but with a different algorithm for resampling, it contains an algorithm for transient recovery, and a very different, intuitive and elegant user interface!

 

 

Do not forget to post a link to working beta. I will be happy to test it and promise to submit my rave (most surely) impressions!

[skerry2006aj]

Seriously? Let's keep discussing this thread's subject.

 

Sorry, i am kind of clueless when it comes to how to advertise my work.

 

The ideea of my works came to me after using HQplayer and liking the SQ a lot, and because i know maths a lot, (I am working at a PhD in mathematics and waves), i thought that it's SQ can be even better if i were to introduce a new algorithm.

[tranz]

Sorry, i am kind of clueless when it comes to how to advertise my work.

 

The ideea of my works came to me after using HQplayer and liking the SQ a lot, and because i know maths a lot, (I am working at a PhD in mathematics and waves), i thought that it's SQ can be even better if i were to introduce a new algorithm.

 

Ok. Let me put it politely. By doing this type of posting, in a competitor's thread no less, you will quickly alienate and irritate those people you are trying to reach. Blowing your trumpet on a bit of vaporware and spamming a bunch of threads is not the way to do this.

 

When you really having something to test or show, setup a dedicated thread asking politely for feedback.