Well they don't work. Just as the other uptone product you have mentioned
Uptone ISO Regen Review and Measurements | Audio Science Review (ASR) Forum
One final thought, then I think I'm done here. I believe I posted earlier, or somewhere else... I manage large audio over IP networks, 100's of sources, 100's of destinations, massive multicast streams. The one area I do see that a switch can affect AoIP is that massive multicast streams can bring a simple-minded unmanaged switch to its knees by flooding it with data bound for 10s or 100's of destinations. When that happens, the streams stop working, and when that happens, there is a very negative audible effect: silence.
I used Enterprise grade Cisco managed switches, like the little one I mentioned above, and it's big brothers, 48 ports +. Long cable runs, some near the maximum for a single Cat5/6 run. Several cascaded switches, and all of this in an extremely hot RF environment. How hot? I sometimes have to loop a Cat5/6 cable through a ferrite doughnut several times to keep the RF from entering a device. I have several of these, the RF causes the ferrite cores to run hot to the touch. That's a LOT of "noise"! If there was going to be an audible switch issue, I'm pretty sure I have the situation that would force it to become audible.
Guess what? We don't drop packets! 100% data integrity. Audio is clean and clear. And to scale this for you, nobody at home is doing multicast streams, much less 100s of net nodes, and nobody at home has the kind of RF noise I deal with every day. If I ain't hearing a problem, neither are you.
Now, to those still holding on tight to the concept of the audibility of a switch, or DIY mods, have you even tried to test your system for data integrity? Dropped packets? Bit-error rate? Don't even know the terms? These are things people in my field look at all the time.
Have you checked for changes to audio performance by measurements of things like distortion, jitter, noise, etc.? If you're not using a managed switch in the first place, you probably have no idea if your packets are good or not, because a dumb switch won't tell you anything. That's partly excusable. But, if you have not even made one single audio measurement to substantiate your claims (the software is free, guys!), and are depending on your personal, biased, subjective opinion, valued as absolute fact, then don't expect an "acceptable" recommendation for a super-audio-switch either. Let's not fall back to "I know what I hear" because it's long been proven that nobody does in the presence of bias. Not me, not anybody.
You want to improve your sound? Find the BIG problems, like the acoustic mess in your room, the performance of your speakers, you know, the non-subtle 20dB+ anomalies, and spend up to fix those. Your net switch is working just fine.
I used Enterprise grade Cisco managed switches, like the little one I mentioned above, and it's big brothers, 48 ports +. Long cable runs, some near the maximum for a single Cat5/6 run. Several cascaded switches, and all of this in an extremely hot RF environment. How hot? I sometimes have to loop a Cat5/6 cable through a ferrite doughnut several times to keep the RF from entering a device. I have several of these, the RF causes the ferrite cores to run hot to the touch. That's a LOT of "noise"! If there was going to be an audible switch issue, I'm pretty sure I have the situation that would force it to become audible.
Guess what? We don't drop packets! 100% data integrity. Audio is clean and clear. And to scale this for you, nobody at home is doing multicast streams, much less 100s of net nodes, and nobody at home has the kind of RF noise I deal with every day. If I ain't hearing a problem, neither are you.
Now, to those still holding on tight to the concept of the audibility of a switch, or DIY mods, have you even tried to test your system for data integrity? Dropped packets? Bit-error rate? Don't even know the terms? These are things people in my field look at all the time.
Have you checked for changes to audio performance by measurements of things like distortion, jitter, noise, etc.? If you're not using a managed switch in the first place, you probably have no idea if your packets are good or not, because a dumb switch won't tell you anything. That's partly excusable. But, if you have not even made one single audio measurement to substantiate your claims (the software is free, guys!), and are depending on your personal, biased, subjective opinion, valued as absolute fact, then don't expect an "acceptable" recommendation for a super-audio-switch either. Let's not fall back to "I know what I hear" because it's long been proven that nobody does in the presence of bias. Not me, not anybody.
You want to improve your sound? Find the BIG problems, like the acoustic mess in your room, the performance of your speakers, you know, the non-subtle 20dB+ anomalies, and spend up to fix those. Your net switch is working just fine.
Btw, here is a little experiment people can try if you are open to the possibility that the switch does matter for sound. Most desktop switches has 5V input and use little power, so buy a battery holder for 4 AA batteries with a 5.5/2.1 DC connector, and 4 batteries, and replace the input power with that. Use it for a few hours and then switch back and see if you notice any difference.
The difference that the above tweak gives is quite small compared to a switch built for audio, but most people with a good HiFi setup and room acoustic will hear an improvement.
The difference that the above tweak gives is quite small compared to a switch built for audio, but most people with a good HiFi setup and room acoustic will hear an improvement.
That test will always pan out as you "want" it to as it is sighted and you know when the change occurs. For both ney and yej sayers. If someone change it an odd day you would never notice it. Why, because there is no difference. And there are no reasonable technical explanations why there should be.
Therefore the troll is you. Keep this to yourself and dont lead others to pay hundreds of dollars for snake oil products.
As you joined here recently and these are one of your first posts its probable that you try to sell something indirectly or lobbying the market. Please stop.
This hobby dont need your ideas. At all. They are humbug. Sorry!
//
Therefore the troll is you. Keep this to yourself and dont lead others to pay hundreds of dollars for snake oil products.
As you joined here recently and these are one of your first posts its probable that you try to sell something indirectly or lobbying the market. Please stop.
This hobby dont need your ideas. At all. They are humbug. Sorry!
//
Last edited:
As TNT noted, the test is fully sighted and biased, therefore invalid. Please note that the problem is that you know what the power system is when you're evaluating it. An ABX/DBT can be done without time limitations, so you could still listen for hours each way, and eliminate bias by introducing the X choice, and by eliminating the knowledge of what X is.
However, all you have to do is use a few tools. Take some audio measurements through the entire system under each power method. See if there's any change. Look at noise, THD, several kinds of IMD, response, presence of jitter, etc. It's very simple to do, anyone with a decent sound card can do this for nothing but the invested time.
But since there's a potential troll here, lets go down this rocky road for a second. "I can hear things that can't be measured!" No, you cannot. The challenge is making the correct measurements. For there to be an audible change, a waveform must change in some aspect. Since all waveform changes are measurable, but may not be obvious, it's important to be thorough. In this case, a change in FR is...well, impossible. But several kinds of distortion could be very revealing, especially two-tone high frequency IMD. Something like 17kHz and 18kHz mixed equally has the potential to be a very difficult signal for many digital systems, but isn't a typical audio test.
The point is, dig in, do the tests, answer the question, methodically, scientifically.
You can learn to compensate for bias, which we all have. For me, listening a day or two and then switching back is usually enough. I know this because I have several times heard what I didn't expect, for example a tweak I did that at first sounded good but later sounded worse than without the tweak.
The problem with measurement is that usually you get a lot of measurement for what we don't hear, and lack some measurement for things that we do hear. But lets take a concrete example:
- Choord Qutest vs RME ADI-2 DAC
They both measure very well and sound very well, so far so good. But while their measurement is very equal, they still sound quite different. Where is the measurement that highlights that difference?
And once you get to a transparent system in a room with good acoustics, soundstage becomes important. What measurement tells the width, height and depth of the soundstage?
And don't get me started with Topping DX3 (which I own myself and use as a TV dac) vs Schiit Yggdrasil. One of those DACs has an open spacious sound, with excellent with, depth and height of soundstage and excellent detail retrieval. The other has a muffled and veiled sound with no soundstage at all and hardly any details. And yet you can't tell that from their measurements, at least not the ones performed by Amir. In fact, the DAC that sound by far worse seems to be better according to his measurements.
So while measurements are good for providing a baseline for how something will sound, they don't tell the whole story (in fact, far from it).
The problem with measurement is that usually you get a lot of measurement for what we don't hear, and lack some measurement for things that we do hear. But lets take a concrete example:
- Choord Qutest vs RME ADI-2 DAC
They both measure very well and sound very well, so far so good. But while their measurement is very equal, they still sound quite different. Where is the measurement that highlights that difference?
And once you get to a transparent system in a room with good acoustics, soundstage becomes important. What measurement tells the width, height and depth of the soundstage?
And don't get me started with Topping DX3 (which I own myself and use as a TV dac) vs Schiit Yggdrasil. One of those DACs has an open spacious sound, with excellent with, depth and height of soundstage and excellent detail retrieval. The other has a muffled and veiled sound with no soundstage at all and hardly any details. And yet you can't tell that from their measurements, at least not the ones performed by Amir. In fact, the DAC that sound by far worse seems to be better according to his measurements.
So while measurements are good for providing a baseline for how something will sound, they don't tell the whole story (in fact, far from it).
The only way to compensate for bias is to eliminate it initially from the test. No human can compensate for bias. This is science, proven, many disciplines.
Completely meaningless. If this were a peer-reviewed paper, you'd be laughed out of the organization.
I'll turn it around: where is the controlled statistical evidence that there is a difference outside of bias?
This is silly. You obviously cannot measure perception qualities. What you absolutely can do is measure aspects, both electrical and acoustic, that impact perception of all of those qualities. It's like wind chill, you can't measure how cold someone feels, but you can measure temperature, humidity, and wind speed.
I must disagree, emphatically. The problem is not the inability of measurements to tell the "whole story", the problem is knowing how measurements relate to the audible. Measurement is now far more sensitive than hearing, but what it cannot do is measure perception. What we know is that when "distortions" (that would include every signal change caused by anything) occur, they can all be measured. Even if you don't know exactly how a measured change is audible, you do know that less change is better.
I hear this anti-measurement argument all of the time. It is advanced universally by those who don't do any measurements, and don't understand the process, much less the application of the resulting data. At one time if fit part of that description myself, back when my measurement capabilities were limited to hand-tuned distortion bridges. I "knew" my measurements didn't tell the whole story. But, we've moved forward 50 years, and now with a set of chirps can collect far more data, and post-process it with far better resolution than was even imagined then. But that also means measurement in general is much harder to understand. It's very easy to dismiss measurement and favor subjective biased listening, but that's a very big trap to fall into, missing the opportunity to improve sound reproduction in a real, tangible way.
- Status
- Not open for further replies.