Multi-bit DSD versus PCM

[Priaptor]

Hi Howie,

The MSB DAC is an amazing DAC no doubt, but you must understand that the R2R section on the MSB DAC, like all R2R DAC's is for PCM only. When you send DSD to the DAC, it's using the DSD section of the DAC. The R2R section isn't being utilized at all. So if you are using the MSB DAC with HQplayer to resample PCM to DSD, that DAC is definitely overkill for the application. You aren't using their R2R section or even the filters in their Sharc DSP chip anymore when using HQplayer. However they do use excellent clocks and have an excellent DSD section, along with well implemented USB interface.

The Merging DAC's use a single Sabre 9008 chip. This chip is an 8 channel chip. In the Hapi/Horus, it's only possible to use each channel separately. There's no internal summing of the channels to improve dynamic range. On the NADAC, the 2 channel version sums all 8 channels to 2 channels internally to improve dynamic range. So this does offer some level of improvement over the 8 channel versions of the the Hapi/Horus and NADAC. However, some folks have built an external summing device to sum all 8 channels to 2 with the Horus/Hapi to achieve the same thing as the 2 channel NADAC does internally.

With any DAC that uses the Sabre chip, it's impossible to completely bypass the multibit SDM section of this chip. HQplayer can allow bypassing of the oversampling filter section, and the modulation, but not the multibit conversion. In my practice, I've found that DAC chips that have the DSDbypass mode sound better with HQplayer using this mode, vs using the mode that passes through the internal SDM/SRC section like the Sabre forces you to only pass through. But this is not saying that a DAC that allows this bypass is going to automatically be the superior DAC. There's much more to the total package than how things are handled in the chip. However if your DAC does have this feature, using it in combination with HQplayer will offer the best performance.

As far as recommending a good DAC. I think you already have a great DAC for the task. It may be overkill for DSD only purposes, but likely handles DSD very well.

Appreciate the answer. Thanks so much

 

[Blizzard]

Appreciate the answer. Thanks so much

No problem

 

[Phelonious Ponk]

If it was inaudible, what would the point of any of this discussion be in this thread? Most people judge audio gear by listening to it. This is the bottom line. You rarely see consumers who are looking for a system show up at the dealer with an audio precision measurement setup to do their auditions. The measurement end of things, is for the manufacturers to use while developing the equipment. It's a very handy tool to have when designing gear. It's a bonus to be able to publish beautiful specs as well, but as you should know by now, the stuff people spend the most on, is the stuff that looks the worst on paper.

So a realistic view should be taken of this. Audio is a subjective thing. These filter/modulators make a subjective difference. But in order to understand exactly what they do to make the difference, you must be very knowledgeable in this area. The other option is to just listen to it and hear for yourself. If it makes no difference to you, then perfect, just keep doing things they way your used to and be happy about it.

Sometimes having a simplistic view on things is the best way to do it. I wish I was capable of this.

This book was a real eye opener on the subject. I don't agree with him 100%, as we need people to do things the hard way. But sure would be stress free to think the way some do. I would fit into the "maximizer" category in his paradigm. My life would be much easier if I was a "satisfier".

This is a question that can be asked over and over in this hobby in which people hear what they want to hear, whether it's there or not, and in which designers and developers fix "problems" lodged deep in the inaudible zone and charge a king's ransom for it. I've demonstrated my own susceptibility to expectation bias to myself, and I'm one of the most skeptical guys on this forum. I trust the ears of those with a strong desire to hear something much less. So when we're talking about stuff that is theoretically unlikely to make an audible difference, confirmed by nothing but audiophile ears, I'll take it with a huge grain of salt. YMMV.

Tim

 

[Blizzard]

This is a question that can be asked over and over in this hobby in which people hear what they want to hear, whether it's there or not, and in which designers and developers fix "problems" lodged deep in the inaudible zone and charge a king's ransom for it. I've demonstrated my own susceptibility to expectation bias to myself, and I'm one of the most skeptical guys on this forum. I trust the ears of those with a strong desire to hear something much less. So when we're talking about stuff that is theoretically unlikely to make an audible difference, confirmed by nothing but audiophile ears, I'll take it with a huge grain of salt. YMMV.

Tim

When it comes to oversampling filter and modulation algorithms, they do make both an audible and measurable difference. This is the main topic on this thread. The great news is the finest of these algorithms possible today (better than even possible on standalone upsamplers such as the dCS vivaldi upsampler) can be preformed with inexpensive software, on inexpensive general purpose computers. So this should be a topic of great interest for you since it will allows to get the best out of inexpensive gear. Not only that you can try it for absolutely free.

 

[esldude]

When it comes to oversampling filter and modulation algorithms, they do make both an audible and measurable difference. This is the main topic on this thread. The great news is the finest of these algorithms possible today (better than even possible on standalone upsamplers such as the dCS vivaldi upsampler) can be preformed with inexpensive software, on inexpensive general purpose computers. So this should be a topic of great interest for you since it will allows to get the best out of inexpensive gear.

Have you confirmed the audibility with something other than open listening though?

 

[Blizzard]

Have you confirmed the audibility with something other than open listening though?

I don't have the measurement gear to do it, but I can assure you that the chip manufacturers who develop these DAC chips have the measurement gear to measure what different modulators and filters do. Do you think a chip like the new ESS 9038pro has 8 different user customizable filter algorithms because all filters sound identical?

Here's an example of some measurements Jussi took from a Teac DAC using his different filters for an example.


http://www.computeraudiophile.com/b...low-frequency-effects-channel-ti-pcm1795-458/

 

[esldude]

I don't have the measurement gear to do it, but I can assure you that the chip manufacturers who develop these DAC chips have the measurement gear to measure what different modulators and filters do. Do you think a chip like the new ESS 9038pro has 8 different user customizable filter algorithms because all filters sound identical?

Here's an example of some measurements Jussi took from a Teac DAC using his different filters for an example.


http://www.computeraudiophile.com/b...low-frequency-effects-channel-ti-pcm1795-458/

Yes, I understand the measurements of real differences. I even have measured that Jussi's software can make a PCM result better with meager gear that I own. You answered the opposite from my question however. While I measured Jussi's result as better I could discern no sound quality difference. Many of these differences while real appear to be at such a low level as to likely be inaudible. So my question was about determining audibility of these improvements.

 

[Muser]

Amusing comment, Phelonious Ponk: "I trust the ears of those with a strong desire to hear something much less."

 

[amirm]

I don't have the measurement gear to do it, but I can assure you that the chip manufacturers who develop these DAC chips have the measurement gear to measure what different modulators and filters do. Do you think a chip like the new ESS 9038pro has 8 different user customizable filter algorithms because all filters sound identical?

Sure. It is called more features on hour data sheet than the next DAC manufacturer.

 

[Blizzard]

Sure. It is called more features on hour data sheet than the next DAC manufacturer.

Yes it's more features, but if there was no audible difference between filters, they wouldn't have them. AKM also has their "Velvet sound" filters. And there's definitely an audible difference with them as well.

 

[Blizzard]

More measured results from the Ifi Micro:


http://www.computeraudiophile.com/blogs/miska/ifi-idsd-micro-measurements-632/

Pretty hard to deny the IMD results between 32/44.1 PCM and HQplayer resampling to DSD 512 on the IFI micro:

 

[amirm]

Here's an example of some measurements Jussi took from a Teac DAC using his different filters for an example.

http://www.computeraudiophile.com/b...low-frequency-effects-channel-ti-pcm1795-458/

Those are very weired measurements. Here is an example:

Look at the X axis. Despite him saying it is a sweep from 0 to 20 Khz, the X axis is 0 to 2.5 Megahertz or 2,500 Khz! Why would I want to see 100X higher bandwidth than hearing range? Using what is there, those spikes appear at around 350 Khz or some 17 times past our hearing limit. Eyeballing what 50 kHz would be, there is only noise there at -105 dbFS.

Second, the caption says he is using a Picoscope. That is a PC based oscilloscope, not an audio analyzer. It comes in 8 to 16 bit analog to digital converters which are insufficient for this type of analysis. No windowing information is provided either. You have to use instrumentation that is far better than the thing you are measuring, i.e. 24-bit ADCs and well designed system with very low analog noise level. Not a scope made for high-speed, low dynamic range as he is using.

I also don't understand why the source signal is at -15 dbFS.

 

[Blizzard]

Those are very weired measurements. Here is an example:

Look at the X axis. Despite him saying it is a sweep from 0 to 20 Khz, the X axis is 0 to 2.5 Megahertz or 2,500 Khz! Why would I want to see 100X higher bandwidth than hearing range? Using what is there, those spikes appear at around 350 Khz or some 17 times past our hearing limit. Eyeballing what 50 kHz would be, there is only noise there at -105 dbFS.

Second, the caption says he is using a Picoscope. That is a PC based oscilloscope, not an audio analyzer. It comes in 8 to 16 bit analog to digital converters which are insufficient for this type of analysis. No windowing information is provided either. You have to use instrumentation that is far better than the thing you are measuring, i.e. 24-bit ADCs and well designed system with very low analog noise level. Not a scope made for high-speed, low dynamic range as he is using.

I also don't understand why the source signal is at -15 dbFS.

Why don't you run some proper tests then? He was likely zoomed in to get a good look at what was going on. I think the test tone IMD results are more important.

 

[amirm]

Yes it's more features, but if there was no audible difference between filters, they wouldn't have them.

Of course they would. It is called more features on the data sheet.

AKM also has their "Velvet sound" filters. And there's definitely an audible difference with them as well.

If there is then a controlled listening test should be easy to run. Is there one?

 

[Blizzard]

Of course they would. It is called more features on the data sheet.


If there is then a controlled listening test should be easy to run. Is there one?

They produce measurably different results as well as subjectively different results. What more do we need? When I toggle through the different filters they sound different to me, when they are measured, they measure differently. That's enough data for me.

 

[amirm]

Why don't you run some proper tests then? He was likely zoomed in to get a good look at what was going on. I think the test tone IMD results are more important.

The King of Audio doesn't take orders from the servants....

 

[amirm]

They produce measurably different results as well as subjectively different results. What more do we need? When I toggle through the different filters they sound different to me, when they are measured, they measure differently. That's enough data for me.

You posts here are to convince others. To that end, controlled listening tests where you don't know which filter is which is needed. Have your wife test you a dozen times without knowing which is which and let us know the percentage right.

 

[Blizzard]

The King of Audio doesn't take orders from the servants....

Sorry king Amir, but it was a question not an order.

 

[esldude]

They produce measurably different results as well as subjectively different results. What more do we need? When I toggle through the different filters they sound different to me, when they are measured, they measure differently. That's enough data for me.

Well Tim's comment earlier expresses my opinion quite well. I know I am susceptible to subjective results not holding up to scrutiny. When that is what others offer I too use a large grain of salt.

So measured differences way down in the noise floors plus subjective confirmation of a difference don't carry very much weight with me.

 

[Blizzard]

You posts here are to convince others. To that end, controlled listening tests where you don't know which filter is which is needed. Have your wife test you a dozen times without knowing which is which and let us know the percentage right.

So far 100% of everyone I've heard who's tried these filters have had positive results. I've had positive results. Measurements show positive results. It also allows me to playback the finest sounding PCM I've heard through my DSD only DAC. Everyone who's heard it so far agrees. This is enough data for me. And for the skeptics, it's a 30 day free trial. If you don't hear a difference, then perfect, it will save you the money of purchasing it. Simply uninstall and go on doing what you were doing before.

Have you tried it yet Amir?