Audiophile Ethernet exists: the AVB/TSN thread

[mppix]

Hi All,

It turns out that audiophile Ethernet exists. It is called Audio Video Bridging (AVB) or Time Sensitive Networking (TSN).
AVB/TSN guarantees latency as opposed to standard Ethernet, which does not guaranetee the delivery or a particular data packet at all. It is an official IEEE standard that is being extended to wireless.

Guarantees can be issued because the protocol is Level 2. This means it requires dedicated switches but they are getting more commonplace.
The protocol has been adopted/promoted by Apple, MOTU, PreSonus, RME, miniDSP, and a number professional speaker/amp companies, e.g. Crown but mostly for professional use cases.

However, the protocol has been largely ignored by audiophiles, which tend to care about bit-perfect audio.

Why?

AVB interfaces available and competitive with non-AVB DACs or interfaces.

Perhaps everybody is waiting for Wireless Time-Sensitive Networking (WTSN)?

 

[Keith_W]

I have heard of AVB, but I do not know what (if any) advantages it has over Ravenna / AES67. Does anyone have a good list of what advantages either Ravenna or AVB have over a standard digital connection, e.g. Toslink, USB, or AES/EBU?

 

[Sal1950]

I hate messing with networking, I always gives me a headache.

 

[alex-z]

For consumers this has little practical use, and in the professional work space Dante already works well.

Sure, there is a licensing cost for Dante, but that never stopped Dolby Vision beating HDR10+, or HEVC beating VP9.

 

[tmtomh]

I don't understand - ethernet works fine already for audio streaming and networking applications, doesn't it? dropped packets are resent, and the timing accuracy of the ethernet connection is irrelevant to the timing accuracy of audio playback.

 

[unpluggged]

I don't understand - ethernet works fine already for audio streaming and networking applications, doesn't it? dropped packets are resent, and the timing accuracy of the ethernet connection is irrelevant to the timing accuracy of audio playback.

AVB is not meant for "audiophile" applications. It's intended for live sound, studio, and broadcast applications with requirements for low-latency and reliable real-time transmission, where general-purpose TCP/IP networks may be inadequate. It's an open standard, but requires software support and specialized networking hardware.

For music consumption, there are no evident advantages over standard TCP/IP networks.

 

[tmtomh]

AVB is not meant for "audiophile" applications. It's intended for live sound, studio, and broadcast applications with requirements for low-latency and reliable real-time transmission, where general-purpose TCP/IP networks may be inadequate. It's an open standard, but requires software support and specialized networking hardware.

For music consumption, there are no evident advantages over standard TCP/IP networks.

Thanks! That's what I thought. Maybe the OP should ask the moderators to change the thread title.

 

[MaxwellsEq]

Big TV and radio studio facilities depend on technology like this. Legacy broadcast technology (especially TV), is driven by master clocks which need to be propagated across the site(s). Live TV and radio, especially interviews between two locations, but also when switching sources requires minimum latency, so deep buffers cannot easily be used*. As a result (unlike domestic setups) clock recovery and synchronization are critical to avoid splats etc. None of this is required in the home.
* I think future implementation may be less dependent on clocks and synchronized protocols

I recommend reading this : https://tech.ebu.ch/files/live/site...9_Patrick_CSq_Live_IP_core_implementation.pdf

 

[somebodyelse]

I have heard of AVB, but I do not know what (if any) advantages it has over Ravenna / AES67. Does anyone have a good list of what advantages either Ravenna or AVB have over a standard digital connection, e.g. Toslink, USB, or AES/EBU?

AVB requires use of standardised but not widely implemented networking QoS features to guarantee end to end bandwidth and latency, even when passing through multiple network switches. The others (Dante, Ravenna, AES67) rely on more widely available but less stringent QoS features that will prioritise traffic but not guarantee bandwidth or latency. The more channels you have, or the bigger the network, the more likely you are to need the guarantee rather than rely on it usually being good enough. I don't think any of them work reliably with WiFi - the Precision Time Protocol they use for clock sync doesn't play well with it.

You should probably be thinking multichannel physical connections like ADAT and MADI rather than Toslink. Network connections work over greater distances, and can be rerouted virtually rather than needing physical connections changing.

The networking capabilities needed for AVB aren't necessarily expensive, but are relatively unusual. You need a network interface with support for hardware timestamping, which you won't find in most PCs or single board computers. There are some inexpensive cards from Intel if you have a spare PCIe slot, and the BeagleBone Black has a compatible network port. Mac owners are in luck too. Otherwise you're looking at an audio interface to act as your gateway to the AVB network. On the network switch side there are some dedicated AVB ones that seem limited and expensive for what they are, but are cheap compared to the enterprise level ones that are suggested for 'proper' installations, or indeed much audiophile HiFi.

 

[Palladium]

I don't understand - ethernet works fine already for audio streaming and networking applications, doesn't it? dropped packets are resent, and the timing accuracy of the ethernet connection is irrelevant to the timing accuracy of audio playback.

I have a hundred or so Cisco switches at work, and so far I have never ever saw one dropping frames on GbE rates despite nearing 10 years of 24/7 continuous operation.

 

[Hayabusa]

Hi All,

It turns out that audiophile Ethernet exists. It is called Audio Video Bridging (AVB) or Time Sensitive Networking (TSN).
AVB/TSN guarantees latency as opposed to standard Ethernet, which does not guaranetee the delivery or a particular data packet at all. It is an official IEEE standard that is being extended to wireless.

Guarantees over wireless? Then the guarantee is it will fail in 1 in so many times?

 

[FrantzM]

I have a hundred or so Cisco switches at work, and so far I have never ever saw one dropping frames on GbE rates despite nearing 10 years of 24/7 continuous operation.

Funny you mention this. We recently decommissioned some switches and looked randomly at their logs on the syslog server ... Not one dropped packet, in a quarter (3 months) on several dozen of GbE interfaces. Not one... Said interfaces had multimedia on them ...
Seriously would like to know what this AVB brings that standard Ethernet doesn't ...
A "solution" in search of a problem?

Peace.

 

[unpluggged]

I have a hundred or so Cisco switches at work, and so far I have never ever saw one dropping frames on GbE rates despite nearing 10 years of 24/7 continuous operation.

It's not about dropping frames (that's being taken care of using buffers). It's about delivering payload reliably at specified timings and in order, which standard Ethernet does not implement (neither it needs to, relying on higher layers to do the job).

 

[Bergante]

Funny you mention this. We recently decommissioned some switches and looked randomly at their logs on the syslog server ... Not one dropped packet, in a quarter (3 months) on several dozen of GbE interfaces. Not one... Said interfaces had multimedia on them ...
Seriously would like to know what this AVB brings that standard Ethernet doesn't ...
A "solution" in search of a problem?

Peace.

The problem is, what happens if you have a busy network combining many workloads? You need the network to prioritise audio traffic properly.

In a big system with many channels and distribution over Ethernet you need clock synchronization.

And, last but not least, all Ethernet interfaces are not created equal. I have been testing Gigabit Ethernet adapters and I found out that the typical Realtek USB-C to GbE can´t sustain 1 Gbps in full duplex, consuming an insane amount of CPU in the process.

 

[tmtomh]

The problem is, what happens if you have a busy network combining many workloads? You need the network to prioritise audio traffic properly.

Okay, but:

1. That has nothing to do with the title of this thread. Not blaming you, but it seems increasingly clear that @mppix should ask for the thread title to be changed because it's totally irrelevant to what they've actually posted about.

2. You pose a question here, but I can't imagine there's not already an answer - or at least data pointing towards an answer - out there. Isn't this one of the main ideas behind ASR, that a good number of the "what if" uncertainties around hi-fi sound actually have some pretty good answers already? A lot of audiophile products are sold based on the typical user not realizing that the product is addressing a nonexistent problem. Again, not saying that you are doing this. Just making a general comment.

 

[Bergante]

Okay, but:

1. That has nothing to do with the title of this thread. Not blaming you, but it seems increasingly clear that @mppix should ask for the thread title to be changed because it's totally irrelevant to what they've actually posted about.

Of course. I was replying to a message whose author wondered wether these pro audio distribution protocols are really a thing or not. They are definitely!

2. You pose a question here, but I can't imagine there's not already an answer - or at least data pointing towards an answer - out there. Isn't this one of the main ideas behind ASR, that a good number of the "what if" uncertainties around hi-fi sound actually have some pretty good answers already? A lot of audiophile products are sold based on the typical user not realizing that the product is addressing a nonexistent problem. Again, not saying that you are doing this. Just making a general comment.

I was not posing a question but trying to illustrate one of the problems addressed by these professional distribution protocols. If your network is not ready to prioritise audio traffic you can suffer drop-outs. But this a problem generally affecting large installations or poor network hardware (I have seen big latency spikes with those crappy GbE adapters).

 

[tmtomh]

Of course. I was replying to a message whose author wondered wether these pro audio distribution protocols are really a thing or not. They are definitely!



I was not posing a question but trying to illustrate one of the problems addressed by these professional distribution protocols. If your network is not ready to prioritise audio traffic you can suffer drop-outs. But this a problem generally affecting large installations or poor network hardware (I have seen big latency spikes with those crappy GbE adapters).

Understood, absolutely - and everything you say makes sense.

 

[kemmler3D]

the protocol has been largely ignored by audiophiles, which tend to care about bit-perfect audio.

Why?

Because "audiophiles" (read: the audiophile press, advertisers, salespeople, and reviewers... arguably all part of the same camp) are simply not interested in real solutions to real problems.

Audiophile network gear addresses fantasy problems, or non-problems, like nanosecond timing of bits filling buffers, or noise levels that don't even cause a failed CRC. They then use the fantasy solutions to fantasy problems to support subjective, nonsense claims like clearer bass or cleaner treble.

AVB/TSN solves a real problem, but it's not one that Audiophiles actually have, and it solves it in a (unsatisfying to them) way that doesn't have to do with imaginary purity of waveforms, so they ignore it.

If they put out AVB/TSN gear they'd have to subordinate their fantasy claims to a real technical body that actually describes how something works in real life, which is not how audiophile writing or reviewing works. Audiophilia cannot survive in a world where problems are actually solved. They can only keep selling ads, and overpriced gear, as long as there is a notion of progress, but not actual solutions. AVB/TSN gear simply solves the problem, end of story. This actually destroys their world.

If they ever introduce AVB/TSN gear, it will be after they identify an imaginary problem with it that can be partially solved with exotic metals, unnecessarily high clock speeds, 3/4" thick aluminum cases, etc. etc.

This is also why "audiophiles" are so bothered by objectivists, who correctly point out that audio electronics are basically a solved problem. This almost literally drives them insane because there's a whole area of the 'hobby' in which there's very little to do.

 

[JeffS7444]

I could see wanting something of the sort for gaming, VOIP or conferencing, but for simple media streaming, I'd think that buffering would suffice..?

My consumer-grade modem/router has settings for media prioritization, but in practice, I saw no real benefits, maybe because traffic on my network is pretty low to begin with.

 

[voodooless]

Hi All,

It turns out that audiophile Ethernet exists. It is called Audio Video Bridging (AVB) or Time Sensitive Networking (TSN).
AVB/TSN guarantees latency as opposed to standard Ethernet, which does not guaranetee the delivery or a particular data packet at all.

Technicality: AVB works on top of Ethernet, just as TCP/IP (or a number of other protocols do), so it’s not an opposing standard at all.