Why does the computer matter?

[Miska]

I'm desperate to know one, seriously.

 

Any good quality pro-audio 8-channel output audio interface with digital room correction and high quality 7.1 channel speaker system playing back 7.1 channel 192/24 FLAC...

 

Unless the listening space is seriously acoustically treated studio.

 

Edit: OK, even with 5.1 channel 96/24 FLAC

 

[barrows]

The prevaling "myth" that I am referring to holds that a simple (and much cheaper) computer based system can outperform very expensive reference quality CD players.

I made no cost considerations in my comparisons, but, when using the PS DL-III (highly modified BTW, and capable of playback performance well above its price) I did make may comparisons using the DAC, ie:

 

bel canto CD-1 as transport in to DL-III vs.

Various computer interfaces. Also into the DL-III.

 

The fireface at over $1000 USD, and the Diverter at well over $1000 USD are certainly not budget items. Notably, optical out of the MacBook equalled or outperformed the Fireface.

 

Of course, my ultimate goal was to achieve computer playback which would clearly outperform the bel canto, and other highly regarded disc players. I have now achieved that with the Wavelink, Pure Music, and my very sophisticated TPA Buffalo II DAC. I expect to take another step up when I elminate SPDIF, when the Twisted Pear USB/I2S interface board is complete and available.

My only point is to debunk the Myth, that still prevails amongst some folks, regarding the general superiority of most (if not all) computer based set ups in comparison with very good CD/Disc players. It takes a very well sorted computer set up to equal or better the performance of some of the very good disc players available.

As Miska notes, there are also the non traditional CD players, like the Memory Player, and PS Audio's PWT, that will always be able to equal any computer based set up, as they are essentially the same thing-but this is a little off the topic.

 

[Windows X]

I tried many pro gears including:

-RME Fireface 400/800

-Weiss Minerva/DAC202/INT202

-Metric Halo ULN-2 (Couldn't get a hand at 8 yet but couldn't perform playback latency below 2-5ms via firewire with 24/192 anyway)

-Prismsound Orpheus

-Lyne AES16e

-RME HDSP AES-32 (Seems to be the best ones for AES/EBU output so far)

-Not to mention numerous shabby USB interfaces to make silver bullet fool-proof products

 

If you can tell the difference between 2ms sound and 5ms sound, you should understand why they're all inferior to $10k or so CD transport...it's not like I'm against this but all I've tried seems to be in vain. And the gap is too high for me to believe Mykerinos can be my savior. I've tried with 0.02ppm masterclock too but not so much luck for winning against Esoteric P-05.

 

I've given my all to become elite computer geek/practical sound engineer/some foolish audiophile who's unlucky enough to realize how good $300k system could be with good setup.

 

[Miska]

I tried many pro gears including

 

My point was...

 

Is the biggest bottleneck of the system; speakers, room acoustics, or any particular flavor of DAC?

 

How about multichannel hires vs stereo CD?

 

From my perspective, it's not black-and-white. If you prefer Esoteric's DA conversion, there's no reason why it couldn't be achieved with computer as a transport, given equal amount of money to spend. Plus the added bonus of having all kinds of things performed with computer not being possible with traditional hardware.

 

[cfmsp]

Well, that was a quick flurry of posts. Ironically, after writing the words, I thought better of it, and I thought I had NOT actually posted it.

 

Couple of comments: I definitely think, as Eloise points out, that any judgments of Computers (as transports) versus dedicated CD transports should be cost comparable (as dCS and others have proven, spare no expense for a given approach and of course, it will best all comers).

 

Ditto for comparisons of full-on CD players versus computers & DACs.

 

Perhaps my experience was atypical (if Barrows is correct in believing that many just assume the computer setup sounds better), but my Metric Halo ULN-2 with properly setup Mac G5 bested my CD player - the not SOTA but still well respected Arcam FMJ CD-23. The costs of the two setups were definitely comparable (although I bought the G5 used) - approx $2000 for each. Substitute a Mac Mini instead of the G5 and the costs remain comparable.

 

It seems to me that perhaps every comment on this thread is potentially correct - i.e. that all out all-in-one players are difficult to best/match with computer-based systems, but perhaps it's a different story on the other end of the price spectrum (where most of the buying activity is)?

 

carry on,

clay

 

 

 

 

 

[Windows X]

Let me correct this. I'm looking for music server to act as transport feeding Esoteric D-05 and the problem is I can't find any good ones that can totally make Esoteric P-05 looks inferior but opposite. I'm not looking for all-in-one solution like CD Player replacement but computer audio transport -> hiend DAC.

 

So far I've heard none to out-class $10k CD Transport via RCA, optical, AES/EBU or dual AES/EBU. Price is no objective if something like Mykerinos can shine better than Esoteric P-05 (or P-03 if possible).

 

If I should talk about CD Player under $5-10k I've already out-classed most of them. They aren't really kind of hiend ones at all. They're a lot of ppl spending $10-15k CD Transport and DAC separately

 

Windows X

 

[Miska]

So far I've heard none to out-class $10k CD Transport via RCA, optical, AES/EBU or dual AES/EBU. Price is no objective if something like Mykerinos can shine better than Esoteric P-05 (or P-03 if possible).

 

Next I would ask if you allow "transport" to also perform heavy DSP? Or what kind of user experience this "transport" is expected to offer? Motion control with 3D graphics?

 

[Windows X]

There're too many substances in computer based system and music server solution to make digital transport becoming as good as $10k CD Transport. If I write them all down, it would take me a long way to explain and ended up in disbelief without real listening experiments. I used to be a guy thinking stuff like decent audio interface with proper configuration can out-class so called highend CD Transport like $10k ones or so. But more I'm experiencing this in higher end system and mine which keeps growing higher and higher, the more I realize about the gap between them.

 

There's also people who experienced WOW effects with DAC202 like this is era of computer audio. When he came to my room and took a few minutes listening between mine and P-05, he admitted mine is a lot better than all he'd heard but P-05 is still way higher to grasp.

 

[audioengr]

Clay wrote"

"How did computers outperform CD Players and CD transports in the first instance (even before/without Async interfaces to DACs)?

 

Are they not able to offer bit-perfect data consistently?

 

Are they more highly jitter-prone?"

 

SOME pre-async USB solutions did have lower jitter than all but the most expensive transports. They also had faster edge-rates and better impedance matching on the S/PDIF interfaces.

 

Some of these won Golden Ear awards and other awards.

 

Steve N.

Empirical Audio

 

[audioengr]

"There's also people who experienced WOW effects with DAC202 like this is era of computer audio. When he came to my room and took a few minutes listening between mine and P-05, he admitted mine is a lot better than all he'd heard but P-05 is still way higher to grasp."

 

I dont doubt this. I have heard from audio clubs in the US where the 202 had varying results depending on the preamp it was driving.

 

Did you try a 202 direct to amps?

 

Steve N.

Empirical Audio

 

[Windows X]

I did and same goes for popular DCS Debussy/Esoteric D-07/Wavelength audio/Ayre/Moon/etc. Almost all popular async and non-async USB dacs are tried and even INT202 won out all of them. It may sound ridiculous but owning decent $100k system can easily catch up a lot of things you normally can't perceive like micro detail, harmonic, transient, etc. Most subtle and hardly noticable becoming a lot clearer once your ears get used to it even in lower-fi systems. I used to think those are placebo effects but they actually aren't.

 

[ItemAudio]

Unspooling the data was never the interesting part of transport design: what makes the Esoteric transports so good is the attention paid to the stuff that matters: power supply and clocking.

 

In the computer domain, this crucial stuff is typically 'outboard' of what many think of as the transport, and all the vital engineering sometimes falls between two stools entirely.

 

Where we are seeing computer audio now getting really interesting is in the plethora of reclocking, buffering, isolation and 'interim devices'. As ever, it's expensive to do this well, but I would maintain that a value equation of some kind still holds true: that pound-for-pound, computers outperform disc transports by a factor of 'X'. The ticket on a P-03 buys a lot of alternative hardware.

 

[The Computer Audiophile]

Steve - Many manufacturers have heard not so great things about your products, yet they don't post it on the internet. Hearing things from audio clubs in the US and reporting it here on the site, when the product directly competes with your products is sketchy at best.

 

Hopefully most of the readers see through most of what you're saying in this and many other posts. Offering "evidence" based on what your customers say is an easy way to say whatever you want on the Internet without taking direct responsibility. As an engineer can you offer opinions based on your own experience and back them up with engineering-like information? That would be great for everyone.

 

[Miska]

DCS Debussy/Esoteric D-07/Wavelength audio/Ayre/Moon/etc

 

How about comparing it this way:

 

1) Expensive transport+DAC, no special processing

2) Reasonably priced but good DAC with a bit of heavy DSP including digital room correction (frequency and step -response correction) on a computer "transport"

 

Otherwise same system, same room.

 

Which one sounds better in an ordinary listening room? At least objectively measured (2) is clearly better in all occasions I've seen[1]. Now if the DAC is replaced in (2) for better one, it gets even better, as in (1), but still (1) cannot reach the level of (2).

 

I'm yet to see a typical listening room without room modes in bass region. It's of course better to fix the acoustics naturally than through processing, and corrections have to be applied carefully, but in many cases there are convenience limits and not everybody can have an acoustically treated dedicated listening room without compromises.

 

So for me personally, loudspeaker listening goes through (luckily mild) corrections and headphones without...

 

 

[1] measured in room: frequency response, step response, waterfall

 

[Purite Audio]

I have been trying a Trinnov St-2 'optimiser' it's an active room correction device, the results in my room have been extremely encouraging, a much larger improvement than and DAC switch,

Keith.

 

[Windows X]

I tried it few minutes ago. Removed master clock with only transport and dac as you speak with same AES/EBU cable. It seems heavy DSP adds delay in sound and make it even worse for purist signal data. Am I biased? No way, the different is too obvious. It's not issue from acoustic or room at all. Plus, I also tested this with headphone system (hooking T1 to Phonitor).

 

I checked your HQPlayer application, sounds promising with introducing some decent DSP filters and 5ms buffer time. Most apps stuck at 50ms and some can reach up to 10ms. However, the listening experiences from each reducing ms below 5ms is quite serious, especially when I managed to make it from 2ms to 1ms and its like I'm listening to linear sound from just 44.1khz.

 

I do agree that non-hiend DAC can reduce the different between music server and CD transport. I could hardly tell the difference from $1k DAC.

 

[The Computer Audiophile]

Hi Windows X - I've been following your recent posts with interest. While your music server experience is frustrating to you it's also frustrating to me. I've heard music servers on the most resolving systems in the world. These servers are the most accurate transports I've ever heard. This is not only my opinion, but the opinion of people who rely on good transports to design equipment further down the chain, like loudspeakers. Also, if you look at what equipment manufacturers are using at audio shows, some of the best are using servers. We all know these manufacturers want the best performance no matter what the transport. They can chose anything in the world yet they chose servers. The physics of a CD transport put it at an incredible disadvantage from the start. A spinning piece of plastic that wobbles is not where I want my data to originate. A completely solid state computer properly configured is a technically better transport.

 

The only transport that can come close to a properly configured server on 44.1 material only is the Spectral SDR-4000S Pro.

 

Also, the discussion about price is a little funny to me. Price doesn't have that much to do with sound quality when it comes to HiFi components and computers. For example Spectral Audio makes the most resolving components in the world, in mine and many others' opinions, yet the price of Spectral components is ridiculously low. I think the Spectral DMA-360 Series 2 power amps for $20,000 may be the best amps available at any price. When I say best I mean most accurate. When it comes to computers price is out the window in my experience. A music server can range from $400 to $80,000. Sound quality does not always go up with price. I can setup a music server, as I've done for some manufacturers and dealers at audio shows (TAD at CES, Magico, Marutani, & Shadi at RMAF) for far less than a $10,000 transport and the performance will much more accurate.

 

I rambled a bit here but the main points are properly configured music servers can be more accurate than any transport, and price has little to do with all of this.

 

 

 

 

 

 

 

 

 

 

 

[cfmsp]

...

 

 

 

[1audio]

"If you can tell the difference between 2ms sound and 5ms sound, you should " be a world class gamer with those reactions, since that represents about 3 feet of travel of sound in air.

 

Why it would make a positive difference defies any analysis of the playback process. 2 mS buffer in the sound chain will require more than twice as many I/O fetches to keep up with the audio stream than a 5 mS buffer and you will always be on the edge of underruns. To make that work well you will need to dig deep into the system. The Merging Technologies Mykerinos solution is to reserve a core in a multicore processor specifically for those processes and prevent Windows from accessing it.

 

Low latency is important for overdubbing in recording but the delay in startup between 2 mS and 20 mS on Windows or a Mac will be swamped by the response of the UI itself. A 20 mS buffer set up correctly will have approx 1/10 the I/O accesses of a 2 mS buffer and consequently a lot less noise inducing activity. Has anyone but me explored longer buffers?

 

There will be confusion between buffers and "memory player". The sound card driver (internal or external) will fetch some data from wherever (ram disk if its a memory player, hard disk if stored locally, network if its a NAS, USB if that is where its stored) and pass it to a buffer in ram (depending on the driver design, there may be two or more buffers). That data is then accessed by the sound card (or USB driver, etc.) and passed to the hardware that pushes it on eventually to the I2S that drives the DAC chips. (All DACs today use I2S to communicate.) Whether or not the system is a memory player the audio passes through this chain (it gets more complicated with interrupt service routines and interrupt priorities etc.). With a "memory player" you need to wait for the entire track to get loaded to ram. In some systems it may actually get pushed back out to virtual memory on a hard disk. And if you load a large file (multichannel at hirez high sample rate for example) you could easily run out of ram.

 

[barrows]

"SOME pre-async USB solutions did have lower jitter than all but the most expensive transports. They also had faster edge-rates and better impedance matching on the S/PDIF interfaces.'

 

While the above is true for many DAC/transport combos, it does not apply to well designed, single box, CD/Disc players. A single box player has many advantages, as I am sure you are well aware: it can be designed with the clock circuit adjacent to the DAC chip, and then the transport mech can be slaved to the DAC clock. The interface between transport and DAC is I2S, avoiding all the inherent jitter increasing problems of SPDIF. A single box player designed this way, with proper engineering, can achieve very low jitter. From a techinical standpoint, there is nothing that would give any computer interface an advantage over this topography, and the end result would only be determined by how well the player is implemented. In high end terms there are moderatley priced CD players which use this approach, the Ayre C-7xeMP for one.

 

 

 

[1audio]

I have spent some time with the Trinnov surround box and can confirm that its performance improvement in room acoustics can be astonishing. You might need to spend a lot more than the $12K for the basic box to get the equivalent improvement with physical changes to a room, some of which may be impossible. Also they tell me that it switches the coefficients on the fly to match the sample rate passing through the box, unlike the others on the market that use sample rate converters at the input.

 

[Windows X]

Thank you for informative opinions, Chris. As for CD Transport reference, I found Esoteric has the world's best CD transport mechanism. Even P-03 surpasses more than half of other products from other companies combined at any price. Let alone P-01.

 

I understand that Mykerinos made very significant shift from sub $1k cards like Lynx and RME (and maybe the rest). I also found music server presents more "accurate" sound than CD transport but remember this. Music server has a lot more chains than CD transport's.

 

CD Transport: laser reading -> transcoding to PCM bitstream/DSD/etc -> encoding to SPDIF/AES -> fetch to output transmitter

 

Music Server: seek from storage (SSD/magnetic/etc.) -> pass through usb/sata interface -> to main bus -> store in RAM (optional and doesn't work for wireless in most cases) -> fetch to CPU's cache and register to process -> fetch output stream back to main bus -> to PCI/USB/Firewire bus -> to audio interface

 

I may miss something though. In CD transport, process is quite simple and less chain to have something to disturb the sound. In music server, there's a lot more than enough to make the sound less resolving, revealing and maintaining all inner details and harmonic from interferences.

 

Please don't get the wrong idea that I'm trying ton imply that music server is inferior. I can setup my system to sound almost as good as DCS Verdi/Puccuni, CEC TL0 or even transport from some $10-20k CD players if not better. You can see my hardware configuration and some information from here http://www.computeraudiophile.com/content/Music-Server-Sound-Quality-Better-Disc#comment-61457

 

But for Esoteric P-03+G-0Rb, it's another story. Even with the same word clock feeding mine won't keep up in their league. Have you tried comparing between Mykerinos system with Esoteric P-03 including G-0Rb clock? I'm curious to hear your impressions. If Mykerinos can work in applications like foobar/J River/etc., I would have higher hopes to try it out some days.

 

[Windows X]

Lower latency can affact output and playback monitoring to a great deal especially in live performance. Many pro users resort to buy decent gears to archive 2ms for live performance and monitoring. 2ms makes 5ms sounds like little lifeless and 1ms is hardly possible for most users.

 

Memory playback is quite troublesome for multi-track hires as you speak. I'm not really into hires files though since I prefer to listen to what I know more than getting great sound from non-interested tracks so most of my tracks are 44.1khz which can play fine from memory. It greatly reduce delay from seeking to magnetic hdd making sound cleaner and clearer.

 

[Miska]

Low latency is important for overdubbing in recording but the delay in startup between 2 mS and 20 mS on Windows or a Mac will be swamped by the response of the UI itself. A 20 mS buffer set up correctly will have approx 1/10 the I/O accesses of a 2 mS buffer and consequently a lot less noise inducing activity. Has anyone but me explored longer buffers?

 

Just to clarify the the buffer sizes in this particular case, I allow number of different buffer size settings between 5 and 250 ms. The exact size may vary due to hardware. I could also offer other values, but I think this was fairly wide range of reasonable values.

 

In HQPlayer case it mostly affects responsiveness of the digital volume control.

 

On suitably tuned Linux it is possible to reliably maintain 1 ms (or even lower) buffer size for full-duplex DSP processing, but there's not much point to have such high overhead in pure player case.

 

[barrows]

for low latency in live monitoring situations, and during the overdubbing process in mixing is clear, but I do not understand windows' emphasis on low latency for music playback? I could care less if the latency is 2 mS, 2nS, or 2 seconds, as long as all the samples arrive at the DAC chip at the correct time relative to each other, latency should have no effect on playback performance. Consider an asynchronous USB interface, the samples are stored in a buffer, and they are clocked out with a fixed frequency oscillator: the timing of these samples will not be altered in any way by something that happens elsewhere in the chain, as the fixed output clock is FIXED, and does not vary to accomodate any other part of the chain.