Odroid or Raspberry Pi2 vs PC for audio

[Qhwoeprktiyns]

Darn, I was right to lose interest in the Raspis then. I still keep my original Raspi Rev B around just so I can use Wolfram Language with it.

 

I believe Olimex cards have much better USB implementation. Will be trying one out later this month and report back.

[Man in a van]

I believe Olimex cards have much better USB implementation. Will be trying one out later this month and report back.

 

 

That's a board that I have missed (there are so many of the dammed things, it's like the twilight zone).

 

Which one ya gettin?

 

 

atb

 

ronnie.

[Qhwoeprktiyns]

that's a board that i have missed (there are so many of the dammed things, it's like the twilight zone).

 

Which one ya gettin?

 

 

Atb

 

ronnie.

 

a10 lime

[Iansr]

- It provides excellent sound from inexpensive DACs.

 

- It allows you to keep up to date with the latest developments just by updating software rather than buying a new DAC.

 

- Yes, it's far easier to customize. A significant part of the "sound" of a DAC is the oversampling and filtering built into the DAC chip or (in fairly expensive models) FPGA. To get different filters, most often you'd need to shop for and buy a different DAC. (A handful of DACs do offer a choice of a couple or even a few filters.) HQPlayer, on the other hand, offers a bunch of filters. You can choose the one you like best, and it's a single software purchase rather than multiple DAC purchases.

 

Those are the reasons I can think of off the top of my head.

All of those reasons appear to be predicated on the assumption that your DAC has shortcomings. So if you are not in that boat . . .

[Iansr]

Not sure about the Raspi 2, but I think all of the previous Raspis had combined USB/Ethernet circuitry - not good for audiophiles.

 

Call me a cynic but this sounds like the kind of thing that could be just audiophile mythology. Is there any persuasive evidence to support it?

[Pepsican]

Call me a cynic but this sounds like the kind of thing that could be just audiophile mythology. Is there any persuasive evidence to support it?

 

Some people heard clicks and pops in the sound. But if you use an expansion card that goes via the I2S port, and you do not use the standard USB ports, you should be fine I think.

[Jud]

All of those reasons appear to be predicated on the assumption that your DAC has shortcomings. So if you are not in that boat . . .

 

All of those reasons are predicated on the assumption you do not have a current Intel CPU in your DAC. In that sense, yes, nearly every consumer DAC on the market has "shortcomings."

 

Edit: They are also predicated on the assumption that if your DAC has anything like the capabilities of the software oversamplers/filters available, it would cost at least an order of magnitude more than the route being proposed here.

[Jud]

Call me a cynic but this sounds like the kind of thing that could be just audiophile mythology.

 

Yeah, I think it's the y axis in Miska's measurements that represents audiophile mythology.

[Jud]

Some people heard clicks and pops in the sound. But if you use an expansion card that goes via the I2S port, and you do not use the standard USB ports, you should be fine I think.

 

Yep. Wish I knew more about I2S than I do.

[Iansr]

Yeah, I think it's the y axis in Miska's measurements that represents audiophile mythology.

 

I have no idea what you are referring to; maybe it is the kind of evidence I was asking about?

 

I use a Pi2 running Volumio, USB out. I have no problems whatsoever. But I was referring to sound quality, not problems.

[Jud]

I have no idea what you are referring to; maybe it is the kind of evidence I was asking about?

 

I use a Pi2 running Volumio, USB out. I have no problems whatsoever. But I was referring to sound quality, not problems.

 

The "problem" is with using the Pi as a Network Audio Adapter (NAA), something for use specifically with the HQPlayer software made by Jussi Laako (Miska). The way this works is that there is a fairly robust computer running HQPlayer, which can be quite computationally intensive. In the NAA arrangement, this computer acts as a server. Then there is an electrically quiet computer, the NAA, that acts as the client, running Miska's Network Audio Daemon, which the server communicates with by means of Ethernet or Wi-Fi. The NAA is in turn connected to the DAC by USB (commonly, though this is not a requirement - Firewire works, and I'm sure there are other possibilities, such as perhaps I2S). So simultaneously using the same bus on the Pi for the incoming Ethernet communication from the server computer and the outgoing USB communication to the DAC may cause problems. If you are *not* using the Pi for an NAA and thus not simultaneously trying to run Ethernet into and USB out of the same bus, no problem.

 

My reference to Miska's measurements is that he likes to do quite a lot of them in his development work, so it's quite possible he would have measurements illustrating the problem with using the Pi as an NAA. Even if he doesn't, Miska is pretty much the last person to engage in "audiophile mythology." He's more the hardheaded engineering type (works for Intel in his day job).

[Iansr]

Jud

re having an intel CPU in the DAC, I'm pretty sure that the designer of the state of the art DAC that I'm in line for would tell you the FPGA can handle all of the processing that is required or desired for best sound quality. And he is undoubtedly one of world's top few DAC designers.

[Iansr]

Thanks for the explanation. HQPlayer sounds like it is similar in principle to JPLAY used in dual PC mode, which I have used in the past.

[Jud]

Jud

re having an intel CPU in the DAC, I'm pretty sure that the designer of the state of the art DAC that I'm in line for would tell you the FPGA can handle all of the processing that is required or desired for best sound quality. And he is undoubtedly one of world's top few DAC designers.

 

I'm sure he would.

 

But I'm guessing said DAC costs considerably more than pairing a general purpose computer you very likely already have with, for example, this SBC with the extra-spiffy 16GB flash drive with Ubuntu pre-installed:

 

ODROID | Hardkernel

 

ODROID | Hardkernel

 

And I'm further guessing that considerable cost differential would remain true after adding the price of HQPlayer.

 

I'm of course not saying folks shouldn't buy "state of the art" DACs. And there's even some future-proofing to a DAC that uses an FPGA, since they generally can be upgraded via firmware. But for people like the OP who are curious about what can be done with SBCs, this is an interesting alternative.

 

BTW, if you don't mind saying, what DAC are you looking at?

 

Edit: P.S. Miska has in the past provided some fairly persuasive explanations as to some of the advantages an Intel CPU may have over an FPGA in terms of price/performance for the specific application of running sample rate conversion and filtering algorithms for digital audio.

[Skip Pack]

An interesting discussion can be found in the following thread:

 

The Best DAC is no DAC - diyAudio

 

which discusses output filtration of a DSD stream in very simple ways. This thread in its later stages gets into the NAA concept, so I think it's germane to our discussion here of taking the RPi2 I2S output for higher data rate DSD.

 

John Swenson has tried/built a output-filter-only DSD DAC and uncovered an unusual distortion that he feels is a deal killer for that simplest of approaches. I believe I read this in a post of his from a couple of years back. He thinks Miska's shift register solution in the DSC1 is the right approach. (I hope I'm stating this correctly.) I have not got my head around the shift register concept at all.

 

Skip

[Iansr]

Jud

well the DAC has been essentially crowd funded and you are right that it will cost more than the HQPlayer solution you have outlined. It will probably go into proper commercial production in due course and then it will doubtless cost considerably more, albeit still a lot cheaper than a lot of other high end DACs. And FPGA firmware updates are a key aspect of future development. The designer is John Westlake. You don't have to read many of John's posts to realise he is 100% straight down the line and incapable of bullshitting.

The thing that puts me off the HQPlayer and the JPLAY dual PC set up is the hassle of having to boot up the PCs when you want to play music, especially if one of them is in another room.

[Pepsican]

I have no idea what you are referring to; maybe it is the kind of evidence I was asking about?

 

I use a Pi2 running Volumio, USB out. I have no problems whatsoever. But I was referring to sound quality, not problems.

 

In your case we're talking probably about streaming Redbook? HQ Player (made by Miska) upsamles to DSD128, which uses a lot more bandwith. Note that Jud said that Miska recommends the rpi could not be used as the NAA in a HQ Player set up. He did not say it could not be used at all.

[amir57bs]

I am looking for a good single board computer , any idea ?

[Qhwoeprktiyns]

Is USB a must ? What DAC do you have ?

 

A RaspberryPi can be a great source, using add-on "hats", such as the Allo Kali Reclocker + SPDIF boards or add-on DACs (so you are not limited to I2S out). Also, in February, Allo will be coming out with a single add-on board for the Raspberry Pi that will provide isolation/reclocking and an SPDIF output (https://community.roonlabs.com/t/allo-digiless-eta/18564).

 

With USB out, you fall prey to adding multiple boxes behind that are much more costly than these cheap hats. Not sure the SQ will be any better.

 

In any case, you will get much better results by running a 2 PC solution: a server (a laptop or cheap itx board can work) and a player (the SBC). You can then use any client/server solution, such as UPNP, Roon, HQ Player, etc...