Dirac and upsampling

[tuga]

A Denafrips Pontus II. You can switch between NOS and OS mode, but don't think the latter is the same as upsampling is it?

Thanks for the advice though. Interesting and useful.

Apologies to all concerned for the thread misdirection.

@karlsushi I have move our conversation to its own thread.

OS is generally upsamplbig.

You could try Dirac and upsample with JRiver to 24/192, or maybe DSD128 if your computer is powerful enough and set the DAC to NOS mode.

 

[tuga]

the Idea is to apply digital room correction in the same processing step as the interpolation / filtering.

And the advantages of doing it with a computer are the higher processing power which allows more complex and efficient filtering and noise shaping, and removing the noise-generated upsampling process from inside the DAC. The DAC becomes just a digital to analogue converter.

 

[karlsushi]

You could try Dirac and upsample with JRiver to 24/192

I'll try this and report back

 

[karlsushi]

Right, an update.

I have changed the settings on JRiver to output at 192kHz irrespective of the source.

I am currently borrowing some ATC actives from an incredibly generous local wammer (separate report to come on that!), so to ensure a level playing field I undertook room measurements again with the current setup.

There is absolutely no doubt in my mind that Dirac is creating some sort of digital distortion right through the frequency range, even when correcting solely on the bass frequencies (under 200Hz).

Switching between Dirac on and off, the filters have the wonderful effect of tightening up the bass and producing a great even response in the low frequencies, but this is clearly at the expense of some sort of digital hardening/distortion/sibilence/dryness, call it what you will, but I'm sure you know what I'm referring to.

Switch Dirac off and everything sounds more natural/vibrant/realistic.

Put simply, Dirac is sucking the life out of the music, which I can only assume is some sort of digital distortion caused by the digital processing.

 

[TheFlash]

a great even response in the low frequencies, but this is clearly at the expense of some sort of digital hardening/distortion/sibilence/dryness, call it what you will, but I'm sure you know what I'm referring to.

Bits being bits…. do you think Dirac might actually be buggering them up in some random way? Or is there some serious analogue noise sneaking in? Or, rather than distortion as most of us would define it, is Dirac actually applying some sort of frequency manipulation in the upper registers despite your configuration of it to affect only the lower?

 

[TheFlash]

Spookily… it would appear you might not be alone.
Post in thread 'At Long Last! My new baby has been delivered'
https://www.hifiwigwam.com/forum/th...ew-baby-has-been-delivered.96470/post-2247028

Scrap that, I’ve just seen your contributions to same.

 

[hearhere]

Put simply, Dirac is sucking the life out of the music, which I can only assume is some sort of digital distortion caused by the digital processing.

I've been trying to draw people's attention to precisely that for a while - see my earlier comments elsewhere eg
https://www.hifiwigwam.com/forum/th...w-baby-has-been-delivered.96470/#post-2247028
The life-sucking features of all these room correction systems (not just Dirac) are inevitable where the entire frequency range is handled by a single full-range amp running an RC processor, even though the processor only ADJUSTS low frequencies. If RC really can't be largely dealt with any other physical means (careful adjustment of speaker positioning, toe-in, tilt, room carpeting, curtains, soft furnishings, etc), then apply Dirac only after the signal has been split by active XO and send only the bass into a bass amp with RC processor, and allow the top end (carrying the "life" of the music) to go unmolested to its own amp and onwards to the top end drivers. Bi-amping can avoid the life-sucking and still benefit the bass where RC may be necessary.

 

[mac72]

I've been trying to draw people's attention to precisely that for a while - see my earlier comments elsewhere eg
[URL]https://www.hifiwigwam.com/for...w-baby-has-been-delivered.96470/#post-2247028[/URL]

The life-sucking features of all these room correction systems (not just Dirac) are inevitable where the entire frequency range is handled by a single full-range amp running an RC processor, even though the processor only ADJUSTS low frequencies. If RC really can't be largely dealt with any other physical means (careful adjustment of speaker positioning, toe-in, tilt, room carpeting, curtains, soft furnishings, etc), then apply Dirac only after the signal has been split by active XO and send only the bass into a bass amp with RC processor, and allow the top end (carrying the "life" of the music) to go unmolested to its own amp and onwards to the top end drivers. Bi-amping can avoid the life-sucking and still benefit the bass where RC may be necessary.

I’m afraid it’s not that simple, for Dirac to work you need to run full frequency sweep but the biggest issue is latency created by A/D and D/A conversion.
I’ve been running bass with a separate amplifier doing basic amplitude correction below 150Hz and I can clearly see the problem in the crossover region - in my case the delay is 1.2mS only , with Dirac it could be even tenfold depending on used device .

 

[rdale]

I've been trying to draw people's attention to precisely that for a while - see my earlier comments elsewhere eg
https://www.hifiwigwam.com/forum/th...w-baby-has-been-delivered.96470/#post-2247028
The life-sucking features of all these room correction systems (not just Dirac) are inevitable where the entire frequency range is handled by a single full-range amp running an RC processor, even though the processor only ADJUSTS low frequencies. If RC really can't be largely dealt with any other physical means (careful adjustment of speaker positioning, toe-in, tilt, room carpeting, curtains, soft furnishings, etc), then apply Dirac only after the signal has been split by active XO and send only the bass into a bass amp with RC processor, and allow the top end (carrying the "life" of the music) to go unmolested to its own amp and onwards to the top end drivers. Bi-amping can avoid the life-sucking and still benefit the bass where RC may be necessary.

I doubt that changing “..room carpeting, curtains, soft furnishings, etc” will have much effect on room modes if that is what you want to fix. You need properly designed and positioned bass traps to do that.

 

[TheFlash]

I believe Karl has no A/D going on because he runs DSP on his laptop with digital source streamed directly onto same device.
However I’m intrigued by this latency issue. I would have expected that if all frequencies went through the same process/processor (as opposed to some frequencies bypassing it entirely) thrn the same latency would apply to all frequencies and it would therefore be a non-thing in sound terms. Am I missing something? Be gentle!

 

[mac72]

I believe Karl has no A/D going on because he runs DSP on his laptop with digital source streamed directly onto same device.
However I’m intrigued by this latency issue. I would have expected that if all frequencies went through the same process/processor (as opposed to some frequencies bypassing it entirely) thrn the same latency would apply to all frequencies and it would therefore be a non-thing in sound terms. Am I missing something? Be gentle!

Yes , if you run whole frequency range through DSP there won’t be any latency issues , this applies to biamping and processing one part of frequency range and leaving the rest “unmolested” ie running bass with DSP.

 

[TheFlash]

Yes , if you run whole frequency range through DSP there won’t be any latency issues , this applies to biamping and processing one part of frequency range and leaving the rest “unmolested” ie running bass with DSP.

Thanks for confirming!

What I then struggle with is the nature of the reported degradation in sound quality through use of DSP. Surely the software has been designed such that the digital signal out is identical to digital signal in where this is what the user has configured. Is it possible to compare pre- and post-processing files above whatever cutoff frequency has been set, and to identify whether there is any difference at all? If the files are identical in terms of 1’s and 0’s, all I can think of (in Karl’s case, where no additional devices are involved and no A/D takes place) is timing. Where adding DSP is done through adding a device, there are of course 2 additional cables and one additional psu to take into account, hence narrowing the question in first instance.

 

[tuga]

According to the developer of HQPlayer "transparent" DSP requires a very high bit depth and also sample rate.
HQPlayer uses at least 64 and up to 128-bit, I think that current AD SHARC modules process at up to 64-bit but the D&D 8C processes internally at only 32-bit/48kHz.

 

[tuga]

Right, an update.

I have changed the settings on JRiver to output at 192kHz irrespective of the source.

I am currently borrowing some ATC actives from an incredibly generous local wammer (separate report to come on that!), so to ensure a level playing field I undertook room measurements again with the current setup.

There is absolutely no doubt in my mind that Dirac is creating some sort of digital distortion right through the frequency range, even when correcting solely on the bass frequencies (under 200Hz).

Switching between Dirac on and off, the filters have the wonderful effect of tightening up the bass and producing a great even response in the low frequencies, but this is clearly at the expense of some sort of digital hardening/distortion/sibilence/dryness, call it what you will, but I'm sure you know what I'm referring to.

Switch Dirac off and everything sounds more natural/vibrant/realistic.

Put simply, Dirac is sucking the life out of the music, which I can only assume is some sort of digital distortion caused by the digital processing.

That is unexpected, and disappointing to hear...

I am using a couple of P-EQ filters for each channel in HQPlayer and turning them on or off does absolutely nothing to the treble or midrange.

 

[tuga]

I’m afraid it’s not that simple, for Dirac to work you need to run full frequency sweep but the biggest issue is latency created by A/D and D/A conversion.
I’ve been running bass with a separate amplifier doing basic amplitude correction below 150Hz and I can clearly see the problem in the crossover region - in my case the delay is 1.2mS only , with Dirac it could be even tenfold depending on used device .

If you are not running the whole signal through ADC then that is to be expected.

 

[mac72]

If you are not running the whole signal through ADC then that is to be expected.

Yes I kind of expected it would create an issue but I hoped pros of fixing room response would outweigh small latency issues . Running whole range through ADC sounds significantly worse even with no PEQ filters applied ,
I’m not giving up yet , for now I reversed to a single amplifier and subwoofer , I’d do bit of listening and measuring then I try again but I don’t hold my breath tbh .

 

[hearhere]

I’m afraid it’s not that simple, for Dirac to work you need to run full frequency sweep but the biggest issue is latency created by A/D and D/A conversion.
I’ve been running bass with a separate amplifier doing basic amplitude correction below 150Hz and I can clearly see the problem in the crossover region - in my case the delay is 1.2mS only , with Dirac it could be even tenfold depending on used device .

No reason for not running a full-frequency sweep, even though the top section is going direct to its own amp and drivers with no DSP. It's the bass that needs attention and the bass amp includes the wherewithall to do this adjustment.

 

[hearhere]

I doubt that changing “..room carpeting, curtains, soft furnishings, etc” will have much effect on room modes if that is what you want to fix. You need properly designed and positioned bass traps to do that.

My very difficult room was dire when I first moved in. It was a new-build with no carpet or curtains and not much by way of soft furnishings. It's a very large almost semi-circular (actually more parabolic) room of 945 sq ft with low ceiling height and floor-to-ceiling glazing on most of the curved wall. These obvious problems have largely been overcome by methods other than electronic DSP room correction to the extent that the sound is better without a Dirac filter. It just needed time, thought and effort to achieve a good sound. This included choosing the right TYPE of speaker, arguably the most important decision that I suggest many people don't properly consider. I bought electrostatcs thinking they may be good, but far from it - they were simply not suitable for my room.

 

[mac72]

No reason for not running a full-frequency sweep, even though the top section is going direct to its own amp and drivers with no DSP. It's the bass that needs attention and the bass amp includes the wherewithall to do this adjustment.

I’m afraid you’re missing the point , if you run part of the signal through DSP there would be delay between signal running with correction and one without . Another fact to consider is the range of frequencies where drivers overlap as you’d end up with a mixture of signal of the same frequencies but some would be delayed in the time .

 

[karlsushi]

My very difficult room was dire when I first moved in. It was a new-build with no carpet or curtains and not much by way of soft furnishings. It's a very large almost semi-circular (actually more parabolic) room of 945 sq ft with low ceiling height and floor-to-ceiling glazing on most of the curved wall. These obvious problems have largely been overcome by methods other than electronic DSP room correction to the extent that the sound is better without a Dirac filter. It just needed time, thought and effort to achieve a good sound. This included choosing the right TYPE of speaker, arguably the most important decision that I suggest many people don't properly consider. I bought electrostatcs thinking they may be good, but far from it - they were simply not suitable for my room.

I believe this may be why you're not actually getting proper bass issues. The issue with bass frequency room modes is created due to parallel walls causing a standing wave bouncing backwards and forwards between the walls. A semi-circular room is actually not a bad room design for audio.