Hq player playlist

[lqbe]

Interesting post. You say Miska "doesn't know much about" the player because he said it doesn't have the DSP features of HQPlayer, then you agree with him it doesn't have those DSP features, but say that's a good thing. So do you also not know much about the player, or is Miska's comment that it doesn't have HQP's DSP features accurate?

 

With regard to DSP being a bad thing if one wants the "purest possible sound quality," that is certainly true if you enjoy listening to 1s and 0s, i.e., the samples that are digital audio with no DSP. (Hint - it's not pretty.) However, if you like most of us would prefer listening to music, then DSP is necessary to convert the digital samples to analog audio you can hear. It's this necessary DSP that Miska's player does very, very well. It is the player you are defending (Resonic Pro) that offers unnecessary DSP designed to change the sound of the converted file, not simply render it in analog, listenable form (for example pitch changes, 'cause y'know we need more Autotune).

Forums and the constant illusion of everyone thinking they know everything. Ok, I'll just stop responding and leave you to it. Besides, reading might help a bit:

 

Pure soundIn order to achieve the purest sound quality, there is absolutely nothing between decoded audio and sound output that might degrade quality, which is essential for both sound and sample previewing and audiophile audio playback.

[bogi]

The more DSP the worse when it comes to purest possible sound quality anyways.

 

You seem not to understand how typical delta sigma DAC works. Otherwise you couldn't write such a sentence.

 

Your DACs aren't pure. They are full of DSP. They do oversampling, delta sigma modulation with noise shaping, some high frequency digital filtering and sometimes also digital volume control. If you think that DAC is 'pure' blackbox and software DSP is something 'wrong', that's incorrect. With PCM sound source and delta sigma DAC architecture there is no such question if DSP can be avoided. The question is only where that DSP will be performed and what quality level it will have. Modern PCs are able to outperform DSP in DAC chips in terms of power, resolution and capabilities. The idea of HQPlayer is to substitute DSP, performed typically in DAC, with software based solution. It depends on capabilities of DAC how much DSP can be performed in software. DACs with digital inputs capable of PCM 384k or DSD512 are about possibility to pair them with suitable SW. Think about price to performance ratio when DSP is done in SW vs. DAC HW.

[Norton]

Modern PCs are able to outperform DSP in DAC chips in terms of power, resolution and capabilities.

 

 

Without necessarily disagreeing with you, it's important to stress that's your opinion, not a matter of consensus. My own experience (using various replay software combined with a DAC with defeatable oversampling and filtering) suggests that the very best results come from letting the DAC do the DSP and using the simplest possible replay software.

 

My electric drill is more powerful than my electric toothbrush, that doesn't mean it's going to do a better job of cleaning my teeth and certainly not without some very undesirable side-effects...

 

Likewise, I find it unlikely that simply using a PC to do the DSP is going to outclass the carefully considered work of DAC designers launching their products into a competitive market. Despite the ubiquity of home computers, I can't for example think of a single vaguely mainstream DAC that's sold on the premise that it has no inbuilt DSP whatsoever as it's intended for use only with a PC running appropriate DSP software.

[Miska]

My electric drill is more powerful than my electric toothbrush, that doesn't mean it's going to do a better job of cleaning my teeth. Likewise, I find it unlikely that simply using a PC to do the DSP is going to outclass the carefully considered work of DAC designers launching their products into a competitive market. I can't for example think of a single mainstream DAC that's sold on the premise that it has no inbuilt DSP whatsoever as it's intended for use only with a PC running appropriate software.

 

Many DACs use COTS DAC chips, costing around $10 per piece. You don't need to look further than their datasheet to realize that the design is constrained by the computational resources available. Another reason is that because it is a mixed-signal IC, there are limits how much computational functionality can be included on the chip before it begins to leak noise to the sensitive analog side a mere millimeter or less away.

 

Due to these resource constraints, typical DACs for example also run proper digital filters only up to 8x (352.8/384k) or 16x (705.6/768) rates and then use ugly approaches like copying the same sample multiple times (SAH/ZOH) or linear interpolation. For the same reason they use low-order modulators that have problems of their own.

 

At higher level the computational resources of a DAC chip are constrained by factors like, just to name a few:

1) TDP - none of the DAC chips I've seen so far require active cooling, so the power consumption has been capped by thermal considerations

2) Single clock domain - all the DAC chips I've seen so far utilize same clock for running both DSP and the conversion section synchronously. This means the clock for the conversion section runs through clock division and also the clock speed is significantly limited, limiting the number of clock cycles available vs input sample. For example typical DAC chip has 512 or 1024 clock cycles per RedBook input sample (22.5792/45.1584 MHz). Compared to i7 6700K processor which runs at 4 GHz and thus has 90703 clock cycles per RedBook input sample to spare, and the processing clock is fully asynchronous vs the conversion clock

3) Limited computational precision - hardware implementations commonly utilize 32-bit fixed point arithmetics causing accuracy errors on complex calculations

4) Limited visibility on input data - DAC chip cannot see data coming later in the song, or contents of the next song. So it cannot do predictive actions or look-ahead analysis of upcoming audio

5) Amount of RAM - DAC chip has very little RAM and you can rarely see separate high speed RAM chips in DACs, certainly not gigabytes

6) Manufacturing process - DAC chips are manufactured in old IC processes, like 120 nm and such, compared to for example 14 nm processes used for modern computer processors, this has huge impact on power efficiency

7) No capability to do digital room correction, active cross-overs or 3D audio for headphones

 

When it comes to DACs, there are some and I'd expect more to come over time.

[bogi]

Without necessarily disagreeing with you, it's important to stress that's your opinion, not a matter of consensus.

 

I agree that everybody's sound preferences are always individual, subjective. Someone can like sound of his DAC DSP and someone else can like more DSP provided by some SW. We own different DACs, we have different sound preferences. Our preferences depend also on what audio equipment, HW and SW we have experience with. Nobody tried and knows 'everything'.

 

But: processing power, resolution and computing capabilities are technical terms, they can be measured or exactly specified, that's not about opinion. Do you think $3 DAC chips outperform $500 PCs in terms of computing power and capabilities for DSP? Do you think computing power and resolution don't affect reachable and measurable quality of DSP? Digital signal processing is based on exact theory and results are measurable. So DSP is not only about subjective opinions.

 

Many members of this forum invested their time in listening comparisons of many possibilities HQPlayer provides. That's the way how to find out what sonic result brings different type and amount of DSP performed either by DAC HW or computer SW. The result always depends on DAC used, but still that's the best way how to learn and understand effects of different digital filters on sound quality.

[Norton]

Do you think $3 DAC chips outperform $500 PCs in terms of computing power and capabilities for DSP?

 

I have no idea, but I do know I prefer the sound of my DAC to a $500 PC doing the DSP. Comparing the trade value of a DAC chip to the retail price of a PC tells nothing about SQ.

 

Many members of this forum invested their time in listening comparisons of many possibilities HQPlayer provides. That's the way how to find out what sonic result brings different type and amount of DSP performed either by DAC HW or computer SW. The result always depends on DAC used, but still that's the best way how to learn and understand effects of different digital filters on sound quality.

 

Yes I'm well aware of that. I've had a HQP licence for 18 months, allowing plenty of time for experimentation in conjunction with a DAC with defeatable upsampling and filters. Heresy I know, but after extensive listening, it's not my replay software of choice.

[bogi]

Comparing the trade value of a DAC chip to the retail price of a PC tells nothing about SQ.

 

Of course if the comparison is done with concrete SW, HW and DAC, it is measurable and then it brings sound quality relevant results like noise level or distortion level. Such concrete measurements can be found for example on Miska's blog.

 

Yes I'm well aware of that. I've had a HQP licence for 18 months, used in conjunction with a DAC with defeatable upsampling and filters. Heresy I know, but after extensive listening it's not my player of choice.

 

Nothing to complain, your way.

[Miska]

I have no idea, but I do know I prefer the sound of my DAC to a $500 PC doing the DSP. Comparing the trade value of a DAC chip to the retail price of a PC tells nothing about SQ.

 

Price of the i7 6700K CPU chip is around $350, if you want to compare chip prices.

 

Of course your subjective perception of the sound quality is yours and nobody can deny that.

 

I'm curious what is your DAC?

[Norton]

Price of the i7 6700K CPU chip is around $350, if you want to compare chip prices.

 

Of course your subjective perception of the sound quality is yours and nobody can deny that.

 

I'm curious what is your DAC?

 

Esoteric K07x.

[Miska]

Esoteric K07x.

 

OK, for that one they've switched from TI to AKM DAC chips and use AK4490. That's 4.56 EUR/piece for small quantities and 2.22 EUR/piece for quantities >1000. Should work quite nice with upsampling to DSD256, especially if the DAC is configured to support the Direct DSD mode of the chip. The DAC chip itself has the typical digital filter to 8x (352.8/384k) rate with stopband attenuation of 100 dB and ripple of +-0.005 dB.

 

My TEAC NT-503 DAC has the same DAC chips.

 

They advertise 34-bit processing, compared to 64-bit floating point processing in HQPlayer.

[beerandmusic]

 

My TEAC NT-503 DAC has the same DAC chips.

 

off topic...since you have the same dac as me, can you tell me what settings you recommend i use on HQPlayer to upsample non-dsd stuff to 4xdsd?

 

And if i use those settings, it won't mess with my desire to play native DSD for my dsd material?

 

e.g.

I want to upsample all pcm but don't want it to touch anything dsd (not even upsample 2x DSD to 4x DSD)...is that possible..can you send screenshot of setttings?

 

Also, can you give me an eta on the release where you will allow click and play without having to scroll to the playlist to play?

[Norton]

There's no way to bypass the internal processing in the AK4490 with either the Esoteric or Teac DAC's. They also go through another layer of their own SRC in FPGA's prior to the processing in the AK4490. When using FPGA's prior to the DAC chips, they must generate a PLL in the FPGA to clock the DAC chip. This adds a pile of jitter compared to clocking direct off the clock pin, or simple divider/prescaler in the digital interface. So with all of that internal processing, I can clearly see why Hqplayer would have little advantage. This would mean 3 stages of SRC before conversion to analog.

 

Really not sure what the point of these posts is. I'm simply reporting my views based on my listening experience. You're not going to change my views by discussion of the cost of DAC chips or citing at best half truths about the operation of my DAC. If you are correct that there is little advantage in using HQP with my DAC, doesn't that undermine the opinions of those who rave about HQP whilst using much the same DAC...

[Miska]

off topic...since you have the same dac as me, can you tell me what settings you recommend i use on HQPlayer to upsample non-dsd stuff to 4xdsd?

 

I use poly-sinc-short-mp (or poly-sinc-short-mp-2s on my Mac Mini because it is lighter on CPU) and ASDM7 modulator. Choice of filter is matter of taste and depends on material. Since I listen mostly rock/pop/jazz/blues that is my choice, but for classical you may want the linear phase variant (without "mp"). Choice between 50 kHz and 150 kHz DSD filter on the DAC doesn't make much difference from objective point of view, so it can be safely selected based on subjective qualities.

 

The DAC chip has to be in the Direct DSD mode, because 150 kHz filter is not available otherwise...

 

And if i use those settings, it won't mess with my desire to play native DSD for my dsd material?

 

The "Direct SDM" check box in DSDIFF/DSF Settings dialog controls that, as long as it is checked, DSD goes through untouched.

 

I want to upsample all pcm but don't want it to touch anything dsd (not even upsample 2x DSD to 4x DSD)...is that possible..can you send screenshot of setttings?

 

Yes, that's the default.

 

Also, can you give me an eta on the release where you will allow click and play without having to scroll to the playlist to play?

 

Hopefully I get 3.14.1 built and tested during this weekend...

[beerandmusic]

^^^ ok thanks...yes, i keep my nt503 in dsd direct...

but if you are saying the default action for non-dsd stuff is to upsample to 4x DSD, why doesn't my teac report DSD on the Display. I rarely play anything but my DSD library anymore, but i am pretty certain that when i play flac files using HQP the teac display still reports 44.1 PCM?

 

Thanks for update on next release.

[Miska]

There's no way to bypass the internal processing in the AK4490 with either the Esoteric or Teac DAC's. They also go through another layer of their own SRC in FPGA's prior to the processing in the AK4490. When using FPGA's prior to the DAC chips, they must generate a PLL in the FPGA to clock the DAC chip. This adds a pile of jitter compared to clocking direct off the clock pin, or simple divider/prescaler in the digital interface. So with all of that internal processing, I can clearly see why Hqplayer would have little advantage. This would mean 3 stages of SRC before conversion to analog.

 

Mmmh, no. See the 'Volume bypass DSDD bit "1"' (DSD Direct) path in the block diagram:

[ATTACH=CONFIG]29606[/ATTACH]

 

"SCF" stands for "Switched Capacitor Filter" which in is analog filter and in this case programmable to two different cut-off frequencies.

[beerandmusic]

Just explaining the reason why you don't notice much of an improvement with Hqplayer and your DAC. The more internal processing you can bypass in your DAC, the better the result you'll have with Hqplayer.

 

I am the least technical inre dacs and audio, so excuse me if i am mis-spoken here, but on the teac, doesn't putting the TEAC in "DSD direct mode" cut off the DACs internal DSD processing and if you feed it HQPlayers upsampled DSD, wouldn't that also cut off the DACS internal DSD processing? (I am assuming when you say processing you mean upsampling?)...the dac will always need to process the stream from one stage to the next..but maybe there is a lot more processing than just the upsampling going on even in direct mode?

[Miska]

That is the AK4490 DAC chip, not the complete block diagram of the digital path inside the Esoteric and Teac DAC's.

 

Sure, but based on my measurements it is accurate regarding the TEAC NT-503 (and very likely for UD-503 too, since it is largely same device). I don't know what the Esoteric does. And the NT-503 has better jitter performance in the DSD256 mode than with PCM input.

 

As a side note, the performance of the built-in DSD upsampling of the TEAC DAC is so horrible that I cannot recommend using it for anything.

[Norton]

Just explaining the reason why you don't notice much of an improvement with Hqplayer and your DAC. The more internal processing you can bypass in your DAC, the better the result you'll have with Hqplayer.

 

The point being precisely that you can configure internal processing or indeed turn it off.

[Miska]

I found the block diagram of the Esoteric K-07X. When you put it in filter bypass mode, you're not bypassing the AK4490 filters, only the Esoteric pre-filtering in their FPGA. Same with Teac. There's no way of using the AK4490's internal DSD bypass mode feature. And even when bypassing the pre-filters, the clock is still being generated via PLL from the FPGA. A true bypass would have relays diverting the path around the SRC FPGA, and putting the AK4490's in bypass mode, but it doesn't. You would also need a different analog filter after the DAC as well, instead of one made for PCM and DSD to do things right.

 

[ATTACH=CONFIG]29607[/ATTACH]

 

Yes, that's on their page. Doesn't have much detail and I don't even know if they ever run the DAC chip in DSD mode or whether it is in direct mode or not.

[Miska]

No, in every possible configuration they use the resource constrained internal SRC/SDM facilities in the AK4490. Bypass mode would require a radically different analog filter after the AK4490's.

 

Maybe, but let's not mix the Esoteric and the TEAC DACs which are different although they use the same DAC chip.

 

Although based on their own block diagram I suspect they use the DAC chip in digital filter bypass mode and have their own digital filters on front. Which limits it to 16x digital filter output rates (705.6/768k).

[Miska]

They both use the same topology. Only the Teac's use cheaper FPGA's with even less processing power for pre-filtering.

 

And UD-503/NT-503 has option to disable the pre-filtering and option to switch the DAC chip's DSD filter frequency (which means it runs in direct mode). The change is also visible in the measured output.

[Miska]

If it does indeed bypass everything, then the analog filter won't be optimized for it. You can't make a optimized analog filter for both DSD bypass mode and non bypass mode.

 

Well, maybe they optimized their analog filter for the bypass mode? Anyway, the DAC performs better with DSD256 input than with typical PCM inputs.

 

An analog filter like that for mixed PCM/DSD use wouldn't be optimum. Especially with SACD's where the noise rises sharply at 20Khz.

 

For that reason, HQPlayer can upsample DSD64 (SACD) to DSD256.

 

 

For most of my listening I use T+A DAC8 DSD in DSD512 mode.

 

P.S. If you want to see optimal DSD DAC, look at my DSC1 DAC design. No DAC chips there.

[hornytoad]

Does upsampling a redbook audio file to Dsd increase the noise associated with playing that file ?

[hornytoad]

Depends on how high you upsample to. DSD 256 takes the noise up to around 120KHz, before it starts rising to levels that would be audible if in the audible band.

 

Great chart here that shows what happens to the noise with PCM and DSD. But only goes up to DSD 128:

 

[ATTACH=CONFIG]29608[/ATTACH]

 

Thanks . So it does increase noise .

 

Upsampling to Dsd is not bit perfect , is it ?

[Miska]

If they optimized the analog filters for bypass mode, it would be poor with all of the other modes that the 99% will be using.

 

Except if they assume them using the built-in DSD256 upsampling...

 

Yes Hqplayer is great for that, but it would be a bonus if the DAC was optimized for DSD 256 rates as well. Optimizing for bypass mode, and DSD 256 rates means sacrificing the sound for those who don't use Hqplayer to upsample all audio to DSD 256.

 

Yes, I know how I would design a DAC if I'd use the same chip. It is not the same way they've done it. But my goal is still to make best out of what ever existing DAC hardware. I take that into account when developing HQPlayer and that's why I have quite a bunch of different DACs and I commonly measure those as part of my testing procedures.

 

The DSC1 would be great if it was available in a turn key case ready to go. Most people don't have the skill to build their own DAC.

 

The design could/will still be improved. And someone could actually manufacture it as long as they conform with the license. But for me, that's not feasible unless someone figures out how to put more than 24 hours in a day.

 

The bottom line is, until we see a DAC designed from ground up to be used with Hqplayer upsampling to DSD 256+, the full potential of your Hqplayer SRC/SDM technology won't be realized.

 

Sure! The T+A DAC8 DSD is probably closest to such at the moment.