The CORE Is Launched

As previously reported, when we announced we were making a high-powered high quality server to play to renderers, many customers contacted us to say they welcomed the high power option to run Roon Server, but they wanted us to add high quality USB audio output to the product. That is not easy to achieve because the electronic noise levels generated by the i3/i5/i7 CPU family have a detrimental effect on the downstream audio quality. But, as it happened, we already had a project working on a solution to that problem running for over a year, and so we accelerated that work and delayed the launch of the CORE. The delay has turned out to be only two months (and 4 days).

The CORE has now been released and is available in two versions.

The CORE Server delivers the original brief to run server applications with ease and play to renderers like the Antipodes EDGE, Roon Ready DACs and other renderers. What sets it apart from using any high power computer is that it is designed to be easy to use, acoustically silent, have very low electronic noise and to use the excellent Antipodes ODAPS external linear power supply to keep switching noise out of your system. The benefits to sound quality of removing noisy steps in the chain are easily heard. Adding to this you get auto-ripping capability and you can easily install your own storage disks.

The CORE Duet variant adds high quality USB Audio output. There is no longer any need to run a separate server and renderer to have it all - the CORE Duet delivers high power and superb sound quality - a dramatic breakthrough. Don't listen to the theoreticians, listen to a CORE instead, and hear for yourself.

CORE Webpage


New Upgrade Paths

CORE+EDGE Solution

The new CORE and EDGE models enable you to pay an entry-level price for true high-end performance for a set of capabilities, and upgrade by extending the range of high-end capabilities over time. At the top of the upgrade path is the CORE+EDGE solution pictured above. This contrasts with paying an entry-level price for entry-level performance and having to trade-up for higher performance. Here is how it works.

Start With An EDGE

Let’s say you have dabbled with computer audio already and either have a music server or run music server software on a Windows or Mac computer or on a NAS. Or perhaps you are new to computer audio and have a computer that can be used as a Server. If so, a first upgrade step might be to buy an EDGE Renderer. Adding an EDGE Renderer provides excellent sound quality at the lowest possible upgrade cost. It means you need to run compatible music server software on your server device, but there is plenty to choose from such as Roon Server, Squeezebox Server, HQPlayer or any DLNA compliant server software.

There are two possible upgrade paths from here. The simplest is to buy a CORE Server, for a CORE+EDGE solution, meaning you run the Server app on the CORE and the Renderer app on the EDGE, and the music is streamed from the CORE to the EDGE over your network. But if you don't need high power (Roon likes high power and managing a large library can require power), you can get excellent audio quality by upgrading your EDGE Renderer to an EDGE Duet. This upgrade is purchased and delivered online, enabling your EDGE to run both Server and Renderer apps. Whether you buy a CORE or upgrade your EDGE, you don't need to send anything to a service centre and you no longer need your old server.

Start With A CORE

You may already have a streamer (a DAC with an Ethernet input), or a third-party renderer. You can add a CORE to use as the server that your streamer or renderer gets its music from. This will improve sound quality, and because the CORE is electronically quiet, you can locate it close to your streamer or renderer and use short high quality Ethernet cabling.

The next upgrade is to use a USB DAC (or the USB input on your streamer if it has one that supports USB Audio 2.0). For this you need to add an EDGE to play from your CORE (the CORE+EDGE solution), or upgrade your CORE Server to CORE Duet. This will provide a dramatic improvement in sound quality. Again this can be done without your CORE Server needing to leave your system.

Upgrade An Existing Antipodes Trio

You may already own an Antipodes Trio, such as the DS Base, DS GT or DX, and want to add high power without losing the excellent sound quality you are already experiencing. All you need to do is purchase a CORE Server to add to your Antipodes Trio and then contact Antipodes Support to book a free session to have your CORE and Trio fully integrated. The integration can be completed online, and the CORE also allows you to add further storage as you need it.

Add Auto-Ripping

If you have an EDGE Duet, CORE Server or CORE Duet, you simply need to buy a basic USB CD/DVD optical drive and attach it via USB to enable auto-ripping. Our software ripping solution will ensure you get the best possible rip by controlling and optimising the speed of ripping You can remove the ripper drive when it is not required.

Roon v1.3 Update

This is an update on Roon v1.3. We can now report that all features of Roon v1.3 work well on all of our current servers, and report a workaround for a small number of DACs that lose sync with Roon under certain circumstances.

An unusual feature of early Roon software is that much of it was single threaded, and therefore did not access all of the power of a server's CPU. This meant something that easily worked well with say MPD would run out of steam in Roon. Roon v1.3 addressed this and Roon runs more efficiently. But the resampling feature in Roon's DSP Engine is still single-threaded and this meant that our servers ran out of steam on resampling on the fly to DSD128. Roon has addressed this by adding a setting they label Parallel Processing that multi-threads the resampling. If you switch this setting to 'on', then our servers have plenty of power to run all DSP Engine features. For example, running the resampling/transcoding to DSD128 on the fly, the processor on our servers sits at only about 35% utilisation.

If you want a high power solution, look no further than the new CORE or the CORE+EDGE solution.

We have encountered one bug with Roon that means that a minority of DACs can lose sync with Roon between tracks when there is a transition between 16 bit and 24 bit tracks and it may happen with other format changes. This does not happen with other player apps and it does not happen with other DACs. It does seem to be a fault in the DAC's USB input, because the Antipodes server is not affected and the situation is only rectified by restarting the DAC. This only affects certain DACs, but it can be readily eliminated by using the DSP Engine Resampling feature to resample files on the fly to a common format.

If you need help with the multi-threading setting in Roon, or the resampling setting to eliminate dropped connections, or want to explore getting an Antipodes solution with very high processing power, please contact Antipodes Support so that we can explain the options.

New High Res Formats

Antipodes has updated the kernel used in its operating system to enable DSD256, DSD512 and PCM 32/768 for a wide range of devices. If you use the Antipodes 2 software suite (on all Antipodes devices and upgraded devices since January 2016) then you should complete a software upgrade. Refer to the 'How To' section for a guide. If you have an Antipodes that was sold with the VortexBox suite, and have only upgraded the software to Antipodes 2 (no hardware upgrade), please contact Antipodes Support before completing this software upgrade, because your unit requires a software patch to be installed in a remote service call (completed over the internet without your Antipodes leaving your system) before completing the upgrade.

You should use a good network connection for this, and you might consider providing a wired Ethernet connection to your router for this particular update, and/or minimising all other internet activity on your network during this upgrade.

We want to add a note of caution about using software to resample/transcode on-the-fly in the server. The beauty of doing it in the server is that the server device can provide you with an ability to make a number of settings that you prefer. But it is questionable whether the server is the best place to do the resampling. For example, resampling a 16 bit 44.1kHz CD rip to 32 bit 384kHz increases the bit-rate by nearly 20 times, and this makes accurate transmission much more difficult. To transmit a digital audio signal with high purity you need much more than the bandwidth matching the bit-rate (since that would only enable a sine wave at that rate). You need a high multiple of the bit-rate to square out the wave to transmit the clock data with any accuracy. So when sending a redbook signal to your DAC, the fidelity of that signal will be greater than when you send a higher bit-rate file. Exactly when the deterioration of the signal becomes audible depends on the cabling and the DAC.

So there is a strong argument for leaving it to the DAC to do the resampling. The downside is that the DAC has to do a lot of calculations to do the resampling, and this can affect the performance of the DAC chip. So it is important to realise there are trade-offs being made, and getting carried away about upsampling to very high rates in the server, just because your DAC can decode those rates, is not wise. As always, let your listening decide what is the best trade-off with your equipment.

Note also that these new high res formats do not have fully specified interoperability standards. The USB Audio 2.0 standard is the only fully specified asynchronous USB standard and only covers PCM to 32bit and 384kHz. Even so, some USB DACs from leading brands do not comply with even this standard. For DSD the situation is even more uncertain because DAC manufacturers are doing their own thing with DSD, particularly with DSD256 and DSD512. This is understandable because there is no fully specified standard for them to build to.

Therefore it is not guaranteed that any Renderer that has DSD playback capability will work with all DACs claiming to have DSD capability. But we can say that we have successfully tested DoP to DSD512 and Native DSD to DSD512 with DACs that have implemented this capability well.

Noone can guarantee what another manufacturer’s equipment can do. Even if we test a DAC from a manufacturer, you might buy the same model of DAC and have problems because of a firmware change by the manufacturer that we don’t know about. Therefore considerable caution is advised with PCM beyond 32 bit and 384kHz, and beyond DSD128. To reiterate, you cannot assume that a DAC that says it can decode DSD256, DSD512 etc will actually decode them except in a range of circumstances that the DAC manufacturer has tested but generally does not tell you about when making claims about the DAC's capabilities. One day we may have a standard that the industry can design to for these high res file formats, but right now it does not look close to becoming a reality. In the meantime, be careful of the hyped claims.

Some customers have asked us why we are not solving this problem from our end. What needs to be understood is that USB audio means that you are literally installing the USB input on your DAC as an audio device in the server - just like if you add a new device like a printer, video card, keyboard etc to your computer. The provider of the device (in this case the DAC) needs to make their device to either work with the native drivers of the operating system used in the server, or provide their proprietary drivers to make their device work. For our part, we are providing the latest Linux native drivers. Unfortunately this means that if the DAC manufacturer has not designed their device to work with Linux native drivers, or if they have not developed proprietary drivers to make their device work on Linux, there is nothing Antipodes can do to address the problem. Our operating system is standard and fully compliant, and it is only within the power of the DAC manufacturers to address the compliance of their DACs.

Of Faith & Science

In many discussions about audio, I often see faith masquerading as science – individuals stating dogmatically what is what and what cannot be, not based on experimentation, but on an untested yet firmly held belief set. In our business we constantly hear firmly held views about ‘but they are just bits’, ‘with Ethernet it does not matter’ etc and the eyes roll back in our heads at the vehemence of the beliefs. The belief set is understandable. The problem is the teenager-like view that they know all there is to know.

The one that I will always remember was on an Australian chat site where an individual exasperatedly wrote, “But for a cable there is only Inductance, Resistance and Capacitance. There is nothing else, unless you can tell me you have discovered a fourth thing, in which case the science community ought to be told (sarcasm).”

I chose this cable example because many audiophiles have discovered that cables do matter after all, so you may see my point. The poster on that site was displaying his faith in a very limited scientific model, and his inability to appreciate that the model is NEVER the same thing as reality. Science does not give us facts, it gives us theories that have been experimentally verified. But the key thing to appreciate is that the limits of the theory are defined by the limits of the experimentation completed.

In order to develop any scientific theory, the first step is to limit the complexity of what is being tested. This is done in two ways.

  1. First you need to decide what variables are relevant and what to ignore. In audio I have heard it a thousand times, where an electrical engineer will state something like ‘at audio frequencies we can ignore that’, and this will be based on an assumption deduced from a theory that may or may not be relevant, and not based on actual experimentation. Deriving knowledge by deduction and assumption is not scientific, but it is common in science.
  2. Second you need to decide what variables are less interesting and are to be held constant, so that the experiment can be focused on a manageable set of variables. For example, in Economics there is a theory about demand curves. This theory studies the relationship between price level and demand level. While any economist will know that there are other influences on demand, such as income levels, information about the product, marketing and distribution, the experiment will hold these constant to expose the influence price has on demand.

So when our colleague stated that ‘there is only Inductance, Resistance and Capacitance’ he was mistakenly thinking that a theory considered good enough for most cable design work was actually a complete description of reality.

This delusion that a scientific theory is somehow an immutable fact or law is entirely unscientific. Real scientists appreciate that even our best theories will only ever be our latest best guess at how reality works. Perhaps more importantly, even if one day our theories do completely describe reality we will not know for certain that this is the case.

And yet this delusion is common. It is common for a good reason. We cannot spend our entire existence wondering about the universe and treating all theories with the scepticism they deserve. We also need to do stuff with what we think we already know, because that is what the science is for. This is what faith is about – choosing to believe something for which we do not have irrefutable proof, for the practical purpose of getting on with our lives. This is not delusion but practicality. We still need to make wires to power our toasters and make the light turn on, so we need to adopt the relevant theories AS IF they were facts. This is where we strike problems. People often have difficulty with adopting faith in the validity of a theory without it leading to dogmatic insistence that the theory is an unchallengeable fact.

High-end audio companies are often criticised for pseudo-science, and often the criticism is fair. Pseudo-science works in marketing and there has been a heck of a lot of it in high-end audio. But high-end audio is also a field where the firms have to push the envelope of what is contemplated within accepted electrical engineering, in order to make better sounding products than their competitors.

Sometimes the scientific debate is because the issues being dealt with are in different scientific fields. For example, we have found that the lattice structure of a metal has a very significant influence on the sound of audio cabling, and this is an issue squarely recognised as relevant in other fields of science but not considered at all in electrical engineering. Changes in the lattice structure during burn in are observable and clearly audible, and yet we cannot detect any change in the measured Inductance, Resistance or Capacitance.

And sometimes it is because the models used in electrical engineering are too simple. For example the main skill in most high-end audio firms is not their invention of new circuits but in their experience, knowledge and skill in the translation of a ‘reference design’ provided by a technology firm into actual circuits. Circuit layout, parts selection, and of course power supplies are often where the differences lie between mid-fi and the high-end. This knowledge is relevant but no, it has not made its way into the text books found at your university yet, and it may never get there.

This is not a phenomenon of just high-end audio. I have had the privilege of working in a range of technology industries and wherever the objective is more than just ‘make it work’ and is instead to ‘optimise how it works’ the knowledge employed goes beyond what is written in text books and is predominantly based on trial and error and the accumulation of insights about what seems to work better.

The next time you hear or read someone insisting on an immutable scientific ‘fact’, please appreciate that he or she is stating their faith and are not representing true science.

And most of all, please appreciate that the opinions I express here are just my humble insights collected over the years in case you are interested. I do not intend to state them as facts and I may very well have experiences in the future that change my mind on any number of these issues. If you came here hoping to find scientific facts about computer audio you will have to seek that delusion elsewhere.

Our Design Approach

Our approach to product development is highly scientific, using blind testing to verify all findings, but we would probably anger many science practitioners with the way we challenge the prevailing scientific theories found in text books. The prevailing science applied in the wider audio industry today is focused on the science of the equipment and fails to address the more complex science of how the brain derives enjoyment from listening to music.

The trouble with applying accepted science to music is we don’t have a reliable objective measurement of how well a piece of equipment conveys the emotion that was conveyed in the original performance. Accepted scientific theories relating to audio fail to experimentally test what counts simply because it cannot be reliably counted, and we will never accept this.

Scientific theories are only accepted when they have been experimentally verified, which means only objective and reliable measurement methods can be used. But this limits the relevance of the theories developed. It is important to appreciate that the limitations of the experiment define the limits of the theory’s relevance. Hundreds of audio chat site debates relate directly to, and yet so often misunderstand, this point. On the one hand someone will argue the existence of a causal effect without any rigorous experimentation to support the view, and someone else will insist the other is deluded since the accepted science does not support there being any such causal link. Both are wrong, but in different ways.

In our view it is a false premise to argue that if the science of the equipment can be perfected then the need to understand how the ear/brain/emotion system works becomes unnecessary. Assuming away the place where all the action occurs is a mistake in a field where we seek to continue to push the envelope of possible performance. Without insights about how we perceive sound and how this leads to our emotional responses to it, the accepted science makes some grossly inaccurate assumptions. For example, the accepted science does bizarre things like reducing the quality of an amplifier to a single harmonic distortion number, or the quality of a digital source to a jitter number. Most audiophiles have learnt the lack of worth of this approach to evaluating amplifiers simply by listening (and thereby measuring what is actually important). A wild-assed guess at the answer to the right question is always more relevant than a precise answer to the wrong one.

Experimentation, to make great audio, has to base its experiments on listeners’ emotional reactions to the music, regardless of how hard it is to measure that. As a result the knowledge gained by high-end audio companies is hard to convey and even harder to prove. In our case, we began by designing servers the way most of our competitors still do - make a simple server and then apply noise-filtering. With years of rigorous testing we have learnt how wrong that approach is. Today we employ no noise filtering and focus exclusively on minimising the generation of noise interference, and just as importantly manipulate the residual noise into what we have discovered are quite benign frequencies. The result is not only sound that eliminates digital harshness and grain, but that renders natural timbre from the inital attack to the tail of the decay. Below are some of our high-level findings to give you an idea of our approach. Antipodes Digital Sources are designed by employing these and other more detailed insights.

  • The mantra that ‘bits is bits’ is easily disproven in listening tests. In reality bits do not exist. Bits are concepts that are represented by electrical, magnetic or physical media and each of these become polluted by distortion to some level in the real world.
  • Distortion of the digital signal is heard in the resultant audio after decoding, even when the bits represented by the signal are not changed.
  • The concept of jitter is a gross simplification of the problem, and springs from the abstraction that the signal is made up of the bits, a clock rate and variation in the clock rate (jitter). This also creates a misleading presumption about the importance of super-accurate clocks. The worth of improving the accuracy of the clock depends on how much noise interference is on the signal that carries the clock information.
  • In a playback system, no amount of noise filtering or buffering/re-clocking used to cure a distorted digital signal is as beneficial to good sound as doing a good job of preventing electronic noise interference with the signal in the first place. This is a key insight that drives our designs.
  • If you increase the bandwidth capability (all else being equal) of the signal transmission, you improve the perceived sound. If you add noise filtering you reduce the bandwidth and damage the impulse response.
  • If you reduce the level (magnitude) of distortion (all else being equal) of the signal transmission, you improve the perceived sound.
  • If you change the frequency rate of the noise distortion of the digital signal (all else being equal), you can significantly change the character and perceived level of the audible distortion after decoding.

Our Apps Approach

Antipodes digital sources use a Fedora Linux operating system designed from the ground up for optimised audio performance. This delivers superior sound to using a Windows or Apple operating system and trying to shed the weight of too many unnecessary processes and a complex audio stack.

Our software is focused on providing optimised audio performance and full back-end integration for third party audio player apps. This means we can not only bring you a range of the best of breed apps that are available on Linux, but also keep bringing them as they emerge. For example, Antipodes delivered Roon on our products within days of Roon Labs offering its first Linux release.

As new killer apps come on the market we can make these available to customers running the Antipodes software suite with just a few mouse clicks and without the unit leaving your audio system.