For the low, low price of $3800, you can be the proud owner of an audiophile grade, 100Base-T network switch: https://innuos.com/phoenixnet/
My PC is very electrically noisy on account of the GPU. This noise was coming out of all the ports, including the Ethernet port. It was making it through that port, by CAT-6 cable, to my switch. The switch then sent it to my Raspberry Pi I was using as a streamer, then into my DAC, amplifier and finally speakers. You could hear it, it was loud when a game started up.
The solution - swap out the CAT-6 with CAT-5 cable which doesn't connect the ground planes. Instantly solved.
I'm considering putting a sacrificial switch from my junk box inline to that one computer so maybe the rest of the network can survive its crashes.
You could verify this by running Wireshark on another machine, or try disable sending pause frames on the machine before it crashes via eg ethtool
In my case just a few weeks ago I had an AP that went bananas, and was directly connected via ethernet to a RPi4, and made the Raspbian kernel panic on the Broadcom driver, randomly. Not long after the AP stopped working completely.
I've also heard it in the 2000s.
There was one song around 1999 that included the sound and also had my ring tone.
It was a very confusing time.
Found it: https://youtu.be/y8ClVMzt9Dw
[edit: added link]
Memcpy can have second order effects due to EMI radiated by CPU, motherboard or coupled via power fluctuations.
There are audible and measurable effects of audio system clock jitter which in turn depends on oscillator power quality.
Someone mentioned that ethernet has isolation transformers. But isolation transformers have parasitic common mode coupling which can inject the noise to ground plane if interference is strong and device is susceptible enough
Watch the EEVBLOG on company which designed scope capable of measuring 1kV common mode signals. (present in EV electronics) https://youtu.be/I7ppDNLlEL4
Audiophile stuff is a weird mix of real effects and snake oil.
If a $10,000 power cable makes your amp sound better to you, who’s to say it doesn’t?
on the other hand, they spent hundreds of dollars for nothing and if they eventually recovered they'd tell other sufferers about it, dragging them down too.
does the victims glimmer of hope justify her scam?
background: she herself suffered from an "incurable" illness for months, where she paid other angel conjurers (that's where she got the idea from). turns out it was just iron deficiency and easily cured with iron supplements. at that point she'd already paid several hundred dollars to those charlatans.
Looking over the product page, it's marked as S/FTP
At least that’s what I was told once upon a time since EE is definitely not my forte.
I'd definitely prefer a DAC with good filtering on its inputs so I wouldn't need to use a middleman, but that is expensive. One with a proper USB module with complete isolation can run ~$5000.
And anti-audiophiles are sometime becoming even worse than manical audiophiles - because of their deep blind belief in their educated superiority (while in reality it's a bit .. untrue).
To me it really highlights the difference of those who seek to understand deeply, and those who interact with the world only in superficial and aesthetic ways. The latter really want to depend on the "magic" within the machine, and can get upset if you try to shed light on how these things really work.
But it’s comically small and solvable much more easily than spending this kind of coin.
I was expecting an AVB switch with locking connectors and redundant power. You know, like a pro audio mission-critical device. Instead it’s basically like using stacks of hundreds to vibration insulate your CD player. It might technically improve things, but the return on investment is poor.
I haven't used a scope, but I have had some laughs plugging things into the same circuit as high gain guitar equipment.
Yes please, let's amplify those audible frequencies to audible levels. A good PSU goes a long way.
https://www.wired.com/2008/06/snake-oil-alert/
(unfortunately this was from Denon who I've had great luck with an entry-ish-level AVR from for years)
> For the low, low price of $3800, you can be the proud owner of an audiophile grade, 100Base-T network switch
That buys you a seriously good piece of switching gear with 10GBe (or even higher) fiber optic links to whatever devices you could possibly want in a home. Heck, even a little over $100 buys you a switch with 4 SFP+ (Mikrotik CRS305).
I guess audiophiles will claim that the optical transceivers introduce latency and _they can hear it_
Jitter on the Ethernet transferring the file will have no effect.
The audio isn't played directly from the ethernet packets. It has to be decompressed, decoded, buffered. In the end of it all the audio goes into a ring buffer and it is played very precisely by the sound card.
The playback itself is a hard-real-time process that needs time guarantees. Failure to fill the ring buffer by any reason can cause audible skips, but not jitter. But most players in practice will simply stop the playback.
You could 100% design a network switch to be good for audio.
- ensure periodic packets like say arp don't introduce jitter
- do qos on audio streams vs other data
- favor latency over throughput
- maybe tune things like MSS to keep packet sizes smaller, but no fragmentation
etc..
For one thing, timing is important. A simple example would be getting video in sync with audio. You could go deeper and manage timing for speakers side to side or front to back. It would get lots harder if a microphone was introduced - then latency would be a big big deal.
These sorts of things are probably why bluetooth isn't used as much with a/v systems.
(lol, here I am arguing the merits of an audiophile network switch)
So long as the amount of audio sent in a packet is larger than how long it takes the next packet to get there, you'll be transfering audio data faster than it is playing, making minor fluctuations in timing between packets irrelevant.
Bluetooth isn't used because BT audio transmission is lossy, audio gets recompressed. Newer BT standards have good quality compression, but you are still recompressing the audio. BT isn't meant to be a high bandwidth protocol.
BT's latency around audio is because 99% of BT implementations suck. Latency can be as low as 50ms or so, but it is often in the 100s or even 200s of ms.
Even major fluctuations. Not talking about the quality here (but the received quality is identical to what is being sent), but just think about Netflix & Co, Imagine if they had to maintain an "ideal network, with no packet loss and constant ping" to your device or otherwise audio and video would be out of sync?
There are protocols that shuffle more or less raw audio streams over the network (Dante for example). In that case yes, you do things to make sure the variables are within a certain range by (usually) segregating the traffic etc, but even then if the timing is off the playback will stop until the stream is reestablished properly. Theoretically it's the same as with any other media stream, just much more sensitive to fluctuations as it is real time (i.e. delay so low you are unable to hear it).
Should you buy an audiophile network switch? God no.
Please don't. You'll make Nyquist sad. https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampli...
On its way to your speakers, a audio file sampled at 192khz will ideally be identical to a recording done at 44.1khz, except over sampling can introduce audio artifacts that degrade audio quality so you can end up with worse audio.
So, please, don't unnecessarily over sample.
The main place you will see this issue is playing audio in something like a video game due to an action like pressing a button, firing a weapon, etc then it can be very noticeable if you're buffering too much audio because it wont sync to what you're seeing and the game will feel laggy even if it's running at a high frame rate.
Excepting subs, it seems many speaker systems are hardwired together downstream from the dac, but there should be a good way to have multiple independent speakers networked together that could be time consistent. and audio <-> video consistent with whatever display you have.
and bnc cables.... would be another whole network.
Is not that every packet is "output" as soon as it arrives to the PC.
Or I didn't understand what are you saying.
A playback device creates a stream of audio and video data that is consistent and together. But if there's a network involved, say you have stereo wifi speakers, then the video and audio streams are separated. Maybe the playback device sends the data to the display using hdmi, and that device sends data over the network to the speakers. Now there are two delays - one for decoding and presenting video, and a second separate one for transmitting and then decoding audio. The second one is variable depending on the network.
But it's delay, and not integrity of the data being played back. If the network sucks and there are retransmissions, your data might still be integrous without loses (depending on the protocol).
It also happens for musicians on stage. They play and after some delay they hear their own sound back. So they use monitors to hear themselves and the rest of the band.
While in lockdown people gathered together to play music with special software. That tells you something about how this problem is well known and how it can be solved in every situation.
Audio data frames are interleaved, ie RLRLRL. There's no point managing timing of network packets because after going over the network, all the data goes into a buffer at the other end before being played.
To do the same with ethernet you would have to use an industrial ethernet implementation of which I know at least three and they are all incompatible with each other and even then you would most likely benefit from an audio focused implementation.
All of this is much harder than just buying a better switch and claiming it improves the sound. Your audio playback device and speaker would have to support the same implementation of industrial internet.
It is a one or zero, it will checksum and TCP either gives the correct packet or drops it. Also I do not know many audio streamers that do not buffer since TCP works way faster than a song can play. Plenty of time to make sure the packets are correct.
I have no idea why the audiophile network switches irritates me the most. At least with audio cables I guess you can do measurements that show the signal is cleaner, even though there is zero chance of you hearing a difference most of the time (I just make sure cables have good shielding and if the cable has to be seen it better look cool too, haha).
Ya know there are even audiophile USB cables.......not a joke google it.
AVB is a set of IEEE standardized 802.1 extensions that provide better control over bandwidth and latency. This is important when you're playing multiple audio and/or video streams to multiple endpoints connected at different points in the network and they need to be low latency and still kept synchronized.
I've seen professional switches from a large vendor where the AVB license cost as much as the rest of the switch.
UDP is another story and it’s normal to hear things.
A patch cable or its connectors would need to be in a dreadful state to get packet loss though, i.e. physically damaged or defective.
For lossless CD-quality audio you need about 160kbyte/s.
So on a gigabit network (125Mbyte/s), you'd need over 99.999% packet loss in order to not keep up.
So they're comparing to an almost completely faulty cable.
I feel like a lot of audiophile stupidity is not understanding scale.
There is a difference between the fancy speaker cables and acceptable cheap ones, but it's so ridiculously minuscule as to be completely inaudible compared to, say, whether or not the listener is wearing socks or not.
And that's not an exaggeration, that difference will literally be more than the difference in cables, probably by an order of magnitude.
The only real difference between the cables will be at frequencies over, say, a GHz. ie RF frequencies 50x higher than cats might be able to hear.
But the audiofool will say, "see, there's a difference - I want my system to sound the best".
And they focus so much on ridiculously overkill cables, amplifiers and stupid stands, yet seem to often ignore room treatment which could really make a difference. It's often massive speakers in horribly reflective minimalistic rooms, driven by amps running at 0.1% of their power rating.
Another thing is not understanding tradeoffs. Everything is a compromise, so when you buy the amp that's rated at 2000W but only run it at 2W, it won't measure as well as the same quality amp that's rated at 50W, also running at 2W.
Similarly a speaker with a max rating of 500W running at 2W won't be as accurate as a speaker with a max rating of 50W running at 2W, because physically the material the speaker is made out of has to be more resilient/less flexible to handle that kind of maximum power.
But they focus on one particular thing (usually the expensive thing) to the detriment of other things.
Also not understanding wattage.
Because sound is logarithmic, it is really eye-opening how little power you need for quite loud audio. Most systems you wouldn't want to listen to at over 5W.
Eg for, say two Kef R7 speakers which have a sensitivity of 88dB, at only 5W, from 3m away is 94.4dB.
That is LOUD.
At 2W, it's 90.5dB. Also LOUD.
https://mehlau.net/audio/spl/ (+6dB for two speakers)
You don't need a 5000W amp. But they don't understand any of this so they just think higher number and more expensive equals better.
It's like buying a Mack truck because you heard it carries more than a hatchback when you're going to the supermarket, and thinking it must be better because it's more expensive.
My FLAC files live on a SSD attached to a Raspberry Pi that serves them up over NFS. The data passes through three gigabit switches before arriving at another Raspberry Pi running Volumio, which is connected to my DAC.
There's enough there to keep an audiophile awake at night, but the bit tests have passed perfectly whenever I've run them. No expensive cables, linear mode power supplies, or audiophile switches would make the tests pass any better.
Storage -> TCP Stream -> Sound output.
However, it never works like this, except maybe in some rare live conditions (in which case, why are you even running this on TCP?). In reality, it goes:
Storage -> TCP Stream -> Ring buffer -> Sound output
Unless you manage to empty your ring buffer (which, in the case of modern hardware can easily be 32+MB, or multiple seconds of audio), you will not hear a single bit of difference. And if you do, it won't come in "after you were meant to hear it", you'll either get a massive gap, or repeats. There is no such thing as "data coming in too late/out of order".
If you’re getting audible errors from your network then literally everything you do on that network would feel like 90s dial up days and with lots of corruption. Most websites wouldn’t even load due to TLS errors.
The problem such arguments as yours is that people have subjective experiences that can be influenced by a multitude of personal emotional states and they then retrofit pseudo-technical explanations for why they perceived those experiences. Or in layman’s terms, you basically just buy an expensive placebo.
The are lots usecases for nearly live audio where TCP stalls cause problems you would never notice browsing the web and would not be a TLS error (which is ushally how networks are 'tested' leaving the problems unfixed).
And before you comment about listening to live performances at home: the broadcaster would be compressing your audio stream anyway. That would be far more impactful to your audio quality than a few dropped networking packets.
Source: I used to work in broadcasting.
Edit: and just to be clear, the biggest reason that live performances are sensitive to latency isn’t because of sound quality. It is because latency is distracting to the performers (eg if their ear piece is ahead of the speakers) and annoying to the viewers if different systems aren’t in sync (like with a movie where the sound is half a second behind the video). So it’s got naff all to do with satisfying audiophiles.
If you want trouble free audio streaming it is worth checking the network.