Gibson Guitars and Ethernet

Gordon_Cabaniss writes "Gibson, the country's second largest guitar manufacturer, teamed up with twelve Silicon Valley engineers and modified the ethernet protocol to link audio between instruments and the mixer. Gibson is calling the technology MAGIC and they are boasting 'both a cleaner sound and a simpler setup.' 'Gibson's Magic carries up to 64 signals per cable, thus saving space and time.' The technology is licensed royalty free and tech giants Sony, Phillips, and Cisco are already showing interest. Gibson also says to not be surprised to see Ethernet ports on guitars within the next 12 to 18 months." I love the idea of my SG having 100mb/s ethernet on it. I'm sure all 3 of my chords would sound ... well, just as bad, but digital.

Comments

mLAN

[wouter]

Yamaha developed a similar technology that could transport audio and midi-signals, going over firewire.

http://www.yamaha.com/proaudio/products/system_m la n.htm

It's an interesting way to hook up sequencers, samplers, synthesizers and sound cards to each other without having to plug in audio and midi wires, and worry about magnetic interference.

mLan can do about 100 separate channels of music (good enough for a Dolby 5.1 system? :) and 16x256 channels of MIDI data. Throughput is up to 200Mbps, so you don't have to worry about MIDI latency again :)

Re:mLAN

[Nater]

Throughput is up to 200Mbps, so you don't have to worry about MIDI latency again :)

Throughput and latency are different things. Here's an example to illustrate:

• In New York, fill a railroad box car with 80GB hard drives filled with your mountain of data (many many many many terabytes in that box car).
• Send the boxcar to San Francisco in the way that one would expect to send a box car from New York to San Franciso.
• Unload the hard drives from the boxcar

Now, some calculations using simple numbers. Let's say you managed to stuff 5 exabytes of data into the box car and it took 3 days to get to San Francisco. Your throughput would be around 34 GB/s. Your latency would be around 3 days.

Is the difference clear now?

Re:mLAN

[Anonymous Coward]

Your analogy may be good but it's sort of wrong-headed. In MIDI, everyone calls the delays that build up "latency", because that's the way it appears to them. You press a key, but don't hear a note right away. It's a functional latency, but it's caused by the little MIDI channel being over-full of data. MIDI was only designed to let one synth control another, not for people to run dozens of channels containing entire scores of music on them. So in this case, increasing throughput decreases what is commonly called "MIDI latency".

Re:mLAN

[dillon_rinker]

Consider ethernet at 10 million bits per second. Those don't all arrive at the end of the second. They arrive in 1500 byte chunks, each of which takes 12 ms to transmit.Yet we do not describe ethernet as having a bandwidth of 1500 bytes/12 ms.

I would conclude that the bandwidth in question is not 5 exabytes/3 days.

Re:mLAN

[StikyPad]

It's not quite as simple as your boxcar analogy.

Here's another example:

• In New York, write the first letter from the first sentence of Marcel Proust's "A La Recherche Du Temps Perdu" on a scrap of paper and tie it around the neck of a pigeon.
• Release the pigeon and wait for him to make the trip to San Francisco and back.
• Repeat with the next letter.

We'll pretend that our pigeon can fly supersonic and requires little or no sleep, so he can make the round trip in under 24 hours. So our actual latency one way would be 12 hours. However, we can't respond until we have transmitted the entire novel. So 9,609,000 days later, the work has been sent in its entirety. Your throughput would be around 1.157407e-5b/s. Your effective latency would be 26 millennia.

--

The Spec

[d5w]

The actual MaGIC spec is available from Gibson's site.

Re:The Gibson Les Paul

[mcj]

I guarantee you that a les paul with this magic stuff built in will cost a lot more than $1500. Hell, $1500 is on the low end of les pauls as it is. :-)

Confused? Maybe this will clear things up.

[SirSlud]

MIDI is not what they are talking about here. They are talking about audio. MIDI is not audio, but rather 'piano roll'. The only data being sent to the sequencer/keyboard is which notes to play, and when. Conversely, audio is the actual audio signal generated by the instrument, whether it be a keyboard or a guitar.

The true value of audio over ethernet is the existing infrastructure (hubs, switchers, etc), coupled with being able to identify 'devices' hooked up to your setup. Mixers, be them software or hardware mixers, that are 'ethernet aware' would be able to auto-assign the devices name that an instrument reports itself as to the network to faders and knobs in your setup. Currently, you have to know which wires are going from which instruments into what audio-ins on your hardware/software mixer/multitrak; in order to fade a guitar line, for instance, you need to be physically aware of which audio-in the guitar is connected to. This amounts to a huge amount of organizational work for producers/techs, as they must use project software or notebooks to keep track of how various projects are wired up. Some technologies are alleviating these troubles, but from what I understand, its still a pain in many setups to keep track of which songs and projects are wired up which way.

Hopefully, this Gibson technology would allow producers and sound guys to forget those details, and just assign 'network instruments' to which ever faders they please, without ever having to verify that the guitar was plugged into the correct audio-in, corresponding to the controls (faders, knobs) you wish you use to do your production and mixdowns.

At least, thats what I get out of it.

BTW, I am a d'n'b producer with a fairly functional grasp of lo-pro to mid-pro MIDI and audio gear, so while I'm not privvy to the nitty gritty of doing sound for live shows or full rack mixers in-studio, I think I can glean what the true pay off is here, for the sound guys and musicians alike.

Re:Sorry, not Ethernet

[Feynman]

I think if you read the spec, you'll find that it conforms to some level of the IEEE 802.3 PHY and link layers. It's the application layer that's been customized.

So, no, it's not "ethernet" in the sense that it complies fully across all layers and clauses of the 802.3 standard for some application. It's simply the same physical and link layers to permit use of standard connectors, media, and other physical-layer hardware.

It is NOT just a new MIDI!

[PastaAnta]

But more along the line of AES/EBU or SPDIF.

MAGIC can transfer digitized sound without (excessive) jitter and latency like AES/EBU. But AES/EBU can only transfer a single stereo channel in 24 bit/92kHz whereas MAGIC can transfer 32 channels in 32 bit/192kHz!

MIDI only transfers information like notes, volume etc. Not the digitized sound itself. This enables you to connect a MIDI keyboard with a synthesizer and produce sound, but NOT to transfer digitized sound from e.g. a guitar to an amplifier or DAT.

The spec. for MAGIC allows for sending control packets containing MIDI information though. Therefore it can ultimately replace BOTH MIDI and AES/EBU etc.

like mLAN?

[version5]

Sounds like Yamaha's mLAN, which is based on IEEE 1394 (Firewire) and carries 100 audio signals, and 4096 MIDI channels, although I think its proprietary. The TechTV website has no mention of clock information being transferred through Gibson's protocol, or even MIDI. I will be very disappointed if they didn't put those into the spec.

If hardware manufacturers actually support this protocol, it will be a huge boon to the home studio hobbyist. Imagine, a 32-in/32-out soundcard for the price of an ethernet card! My money is on Yamaha. They already have hardware that supports their standard, including a couple of digital mixers. If only Gibson and Yamaha would work together on this, we might have a slight hope of interoperable standards.

Re:will musicians buy this?

[given_to_fly]

As a guitarist my self, I can pretty much agree with you. I , and most of the guitarists i know, are into analog circuit effects an tube powered amps. There is just more warmth.. its not as clean.

however there are quite a few of dsp's that simulate vintage amps.. Line 6's POD and amps, Fender Cyber twin.

while these do sound pretty impressive and do give some extra flexibility, they still don't match a Class A amp.

I could not see any musicina in the near future pluging cat 5 into their guitar.. maybe into a set of mic pres that then went to the mixer to clean the mass of cables up .. or for midi and such.

just my thoughts.

Re:Sorry, not Ethernet

[anticypher]

Did you read the spec? Looks like ethernet to me.

At the physical layer, they have chosen to use inline power ethernet, an emerging standard. Data pins remain unchanged. Power over ethernet seems to be optional, its just there for unpowered devices like acoustic guitars.

At the link layer, they conform to standard MAC addressing, and leave space for IP/TCP/UDP headers, so the signals can be routed/bridged.

There are new packet definitions for timing and other functions, but I wouldn't be surprised if I could just plug a pair of cisco routers in between some of these devices and make them work across a town. It might take some careful bridging configuration, but it looks like straight forward networking at layer 2.

The next higher layer jumps straight to application layer(7), and defines audio channels and control signals. As a networking person, I couldn't care less about that, I'll deliver any packet payload. And the application doesn't really care whether I moved the signals over fibre or copper or a WAN link.

Given their careful stepping around of the IP/TCP header area, I'd say when these devices exist, they will bridge/repeat any other IP traffic that obviously isn't MAGIC traffic. So you can have a browser behind your mix panel with only a single connection to the local router, and your friends can be playing their instruments behind their DSL connections.

For geek factor, I'd give this a 9. Very cool.

I'll leave the "gouging the musicians" comments to the musicians :-)

the AC

Yes, but mLAN *is* FireWire

[s.o.terica]

Why would anyone use this when they could use Yamaha's mLAN anyway? mLAN's based on FireWire, so it's much faster and has the advantage of having a built-in isochoronous (time-dependent) transport protocol.
It's clearly the audio bus of the future (due in no small part to the fact that it can be connected to most off-the-shelf computers these days) -- it's even already supported in Mac OS X Core Audio.

Re:Technically superior?

[Just Some Guy]

If your final format is CD, then won't you loose the same "coloration" of the sound if your guitar/keyboard/mixer is digital?

The idea is that each component that modifies the digital signal slightly degrades the sound quality. If you turn the volume down on a digital mixer, then you're effectively throwing away the least significant bits of the audio stream. The goal, then, is to sample your sound at a high enough resolution that even after all of those lossy transformations, you end up with enough bits left over to fill the CD master's dynamic range.