MADI or Dante?

[Rich Walsh]

If you were specifying a brand new system from the ground up, which way would you go these days?

My gut feeling at the moment is that I’d throw Dante into a rehearsal situation, but tend towards MADI for a production system – partly due to integration with DiGiCo. I’m also slightly wary of the fact that Dante seems to require a knowledge of IP systems that is beyond most sound engineers at the moment, whilst MADI is basically AES-EBU on a connector you have to crimp rather than solder.

Grateful for any thoughts while I ponder this…

Thanks.

Rich

[George Wirges]

I’ve used both, and both are pretty straight forward. I lean more towards Dante mostly because I’ve gotten used to it. Really doesn’t require that much IP knowledge (although my systems are static IP) since it can be used DHCP, so it is basically ‘plug-n-play’

My 2 cents

George Wirges
Wirges Sound Designs
“Just Listen”

Sent from my iPhone

> --
> Contact support anytime: sup...@figure53.com
> Follow Figure 53 on Twitter: https://twitter.com/Figure53
> User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
> ---
> You received this message because you are subscribed to the Google Groups "QLab" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/ACC4CC75-2DBC-43A1-A47B-E88B5829D7ED%40mac.com.
> For more options, visit https://groups.google.com/d/optout.

[Brendan Aanes]

Is your question about what to use for playback, or a venue-wide audio distribution backbone?

If the latter, to me this is somewhat of an apples to oranges comparison and the use cases don’t totally overlap. Certainly Dante can do what MADI does as a single-wire multichannel transmission scheme, but having devices on a network means there are vastly more possibilities for routing, which open up a multitude of use cases that in my shows I’m still discovering. For example, setting up an A2 monitoring station can happen on a computer running DVS and some routing software (or with one of the newer 2ch Dante headphone amps). System tuning can all be done with DVS and a Ri-8, or just spare console inputs for measurement mics. Need to record something in tech from the band or an actor? Just send it to the tech table via dante.

But, if MADI covers all of your possible use cases and you don’t forsee any unusual routing being useful in the future, then yes, it is a more straightforward protocol. I do think a well-designed Dante network can be highly reliable and not require IP knowledge from the daily users as long as they are given basic guidelines on what not to do, and maybe what to reboot when if there is a failure. The one point of failure I don’t like is that DVS exists only the primary network, but that can be solved by using Yamaha PCIe cards (in a thunderbolt chassis if necessary) or their counterpart from Focusrite.

One other point is that latency will be higher on a Dante network than with MADI, but Dante network latency is typically 1ms or 0.5 with a smaller network, which is largely irrelevant unless you’re hopping between a lot of devices. As far as I know MADI connection latency is essentially zero if no SRC is involved.

> To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/DD7C8333-D60A-4E80-894F-96D256CCE2AD%40yahoo.com.

[Rich Walsh]

Specifically about a QLab playback system – in fact, a dual-redundant one. Also, for a venue rather than a specific show or set of users.

Dante is only plug & play until it _doesn’t_ work – at which point I think it is going to be harder to troubleshoot. I think you’re right that if you don’t need the infinite routing (and breakability) of Dante, MADI is a more robust solution – if you can afford it. A well-designed system is going to be simpler in MADI. I think it’s fine to not run a secondary network from your playback, in general.

Last time I did this exercise we were coming out on ADAT Optical, and then converting to MADI. 5 years before that it was ADAT Optical to analogue.

Rich

[Johnson, Philip]

I’d go with Dante. I haven’t found many issues with having to know a lot about IT.  It’s plug and play mostly and with audinate’s via software any computer can be an input or output. I wish we would have specified Dante on our other system which uses madi.  Both work great but I’m liking Dante more and more. 

Sent from my iPhone

On Nov 21, 2017, at 8:56 AM, Rich Walsh <rich...@mac.com> wrote:

[Rich Walsh]

I think this gets to the nub of my concerns (I am very familiar with both protocols, and AVB): give someone a system with an RJ45 on it and somewhere down the line it’s going to end up on a network that isn’t "plug and play mostly” and requires proper system design – just like wordclock or phase coherence or gain structure – that is currently beyond the abilities (or even awareness) of a lot of people who work in audio. See the recent thread about backup switches. Someone needs to think about IP addresses, broadcast traffic, network security, redundancy, failover, backups, etc…

It’s a bit like QLab: it’s very easy to get it basically working and do quite a lot with it, but as soon as you want to do more complex things you’ve got to learn how it actually works. Or look at Mic’s Cookbook.

Rich

[micpool]

It's hard to say because Dante changeovers for dual redundant playback systems haven't been around as long as MADI changeovers.

Using 2 Digico UB Madi Interfaces with a changeover works well and because they were available before commonly available Dante changeovers you often find them as the QLab interface even when the rest of the system is Dante.

If you are interfacing with modern Yamaha consoles MADI is going to knock out 2 card slots if you want 32 and isn't an option for  48 channels if you only have 2 slots, whereas you get 64 channels of Dante with the built in connector. The new Sony and Shure Wireless microphones have Dante options

Mic

[Charles Coes Lists]

Depends entirely on the desk you're putting it into - SD series Digico? MADI all the way - especially if you are optocore for stage racks - changeover is as easy as A/B inputs.... Yamaha CL series - Dante is all over the place... New Avid Desk? AVB...

Dante can be easy to setup and recall, and has better troubleshooting tools built in than MADI, and is Bi-Directional for when you need to grab things for B-roll or virtual sound check, and clock is a much simpler thing to solve if it goes wonky...

MADI works mostly when you plug it in, and usually when it isn't working, it's because you plugged it in wrong, or clocked it wrong...


Charles Coes
cco...@gmail.com
"Words ought to be a little wild, for they are the assault of thoughts on the unthinking" -John Maynard Keynes

[John Huntington]

I don't have a lot of time tonight to address a bunch of the issues but:

On Tuesday, November 21, 2017 at 6:24:49 PM UTC-5, Charles Coes wrote:

Depends entirely on the desk you're putting it into -

Hear, hear.

However, I would say there are a lot of misconceptions in this thread about Dante that can easily be dispelled by doing the Audinate Level 1 training, which is free online:

https://www.audinate.com/resources/training-and-tutorials/dante-certification-training

The troubleshooting tools in the recent versions of Dante Controller are pretty good and the testing approach taught in Dante's Level 2 have really advanced things.  

Also, regarding AVB, Roland Hemming posted this just recently, and I while I haven't had time to read through this carefully on a quick read I mostly concur:

http://rhconsulting.eu/blog/files/AVB.html

Audinate has funded some of Roland's research but he is definitely independent.

And of course I've written thousands of words about Dante and AVB on my blog:

http://controlgeek.net/blog/?category=AVB

John

[micpool]

I think that anyone who is using Dante on a regular basis should certify to Level 2.

I did level 1 on line at 6am,  then went to a 1 day level 2 training at Plasa, and did the level 2 test on line on the train on the way home.

It's a well designed course and covers the basics and a lot of stuff that isn't immediately apparent.

Level 3 is for people who want to wire up football stadiums or international multi site broadcasters.

Mic

[Rusty Wandall]

The new SD12 has DMI card options that can give you the best of both MADI and Dante on the console. Dante has a lot flexibility as others have mentioned. MADI is simpler to set up. However, Dante does have a lot of ways it can be configured with little networking knowledge and a free training and certification program online.

You could also consider an Orange Box or other MADI-Dante bridge if you are wanting Dante on a non SD-12. With the amount of gear coming out wirh Dante, it would be hard to stay away from it completely.

Rusty

[ric...@rbisound.com]

I would go so far as to say no matter if you use Dante on a regular basis or not, you should know your way around it if you want to work in our industry.  I have not gotten the certifications yet but I found Dante simple and easy to set up.  The only thing that I would consider outside of basic computer network knowledge (and I would say mandatory knowledge to work in pro audio) is setting up the switch to prioritize packets.  And realistically you probably only need that in more complex networks.

I think if you're not up to speed on the basics concepts in an IT network at least up to what is required to set up Dante you are going to find yourself left behind technologically very soon, if not already.  Understanding basic IT infrastructure is just going to be mandatory skills for pro audio, just like understanding balanced audio, ground loops, basics of digital audio, etc..

Richard

Get Outlook for Android

The new SD12 has DMI card options that can give you the best of both MADI and Dante on the console. Dante has a lot flexibility as others have mentioned. MADI is simpler to set up. However, Dante does have a lot of ways it can be configured with little networking knowledge and a free training and certification program online. 

You could also consider an Orange Box or other MADI-Dante bridge if you are wanting Dante on a non SD-12.  With the amount of gear coming out wirh Dante, it would be hard to stay away from it completely. 

Rusty

-- 
Contact support anytime: sup...@figure53.com

Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
--- 
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com

.
To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/79fb7654-1109-4ff6-9968-57092097c04c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[John Huntington]

On 11/21/2017 10:17 PM, ric...@rbisound.com wrote:

The only thing that I would consider outside of basic computer network knowledge (and I would say mandatory knowledge to work in pro audio) is setting up the switch to prioritize packets.  And realistically you probably only need that in more complex networks.

Audinate's official position on this is if it works, don't mess with it :-)  You really only need QoS if you're sharing your network with a regular IT network or something like that.  Unmanaged (and non EEE) switches are fine for the vast majority of applications.

I think if you're not up to speed on the basics concepts in an IT network at least up to what is required to set up Dante you are going to find yourself left behind technologically very soon, if not already.  Understanding basic IT infrastructure is just going to be mandatory skills for pro audio, just like understanding balanced audio, ground loops, basics of digital audio, etc..

I agree... We require my basic control systems/networking class of all our students.  So they've at least been taught the information.  Whether they retain it or not... I can't guarantee it :-)

John

[Rich Walsh]

I’ve slightly lost track of which comments are directed at who, but:

• I can’t enforce that end users of a system I spec take a course to know how to use it. I agree that if you take your job seriously you should understand Dante these days, and by association you should know a fair amount about networking. Having said that, I have worked with very senior people at major theatres who can not solder, put a 13A plug on properly, set gain to use more than 2 bits of Pro Tools (genuinely: I did have to troubleshoot a recording done on a U87 that wasn’t even metering at the inputs), tell if something’s using DHCP or not, and so on…
• I disagree that there are misconceptions about Dante in this thread. If you can cite a specific statement about Dante that is wrong then I would like to be educated. I reiterate the fact that if you give someone a system with an RJ45 on it, somewhere down the line that is going to end up plugged into a wider network and somebody needs to think about that. I fundamentally disagree that Dante is “simple”. It is “simple” to plug a mic into a powered speaker and make noise; that doesn’t make analogue audio systems or their design “simple”.
• Thanks to Mic and Charles who made the really valid point that the best approach is to pick the format that integrates best with the particular desk you are dealing with (for now). Dante into DiGiCo is mildly irritating – and you need to allow for any DiGiCo, not just the latest SD12 that is moving towards a simpler Dante integration. MADI into Yamaha is intensely irritating, on the other hand. Either way you end up buying lots of extra kit if you don’t come out of the computer in the flavour the desk likes…

In terms of answering the question I actually asked, is the consensus is that you should pick the format that best matches the desk and then mitigate accordingly – eg: include a gateway to the corporate network at the design stage and provide Audinate training for the current users, or be sure to label the BNCs so they get plugged into the right holes (you’re also going to need to think about a wordclock master either way)?

Now, as to whether your whole system should be an Optocore loop or a Dante network – given it’s increasing ubiquity – that’s a different question that I didn’t ask… Hmm, I’d have to weigh up flexibility against how easily someone could break it, and whether I’d trust them to know when to stop fiddling because they were out of their depth.

Rich

[John Huntington]

On 11/22/2017 6:26 AM, Rich Walsh wrote:

I’ve slightly lost track of which comments are directed at who, but:

• I can’t enforce that end users of a system I spec take a course to know how to use it.

Do you really think your users can step up and route MADI through/to a Digico (or CL5, or whatever) without some sort of training or assistance?  Training/education is a reality of every system today.  At least Audinate has a pretty well done course available online for free.  I recommend it for everyone because even if Dante goes away tomorrow (unlikely) networked audio is here to stay.  People need to learn it.

• I disagree that there are misconceptions about Dante in this thread. If you can cite a specific statement about Dante that is wrong then I would like to be educated.

I would start with your first statement, "... Dante seems to require a knowledge of IP systems that is beyond most sound engineers at the moment, " ; also people subsequently talked about using DHCP or fixed IP addresses for Dante. Audinate now recommends in their training that you just use link-local addresses (automatically assigned without a DHCP server), and this means for 90% of systems (and 100% of small, stand-alone systems) if you use an unmanaged switch you really need to know very little about networking--it really is plug and play.  Even on larger networks you don't need to do much.  We run the Gravesend Inn on a managed network with four switches and 7 vlans and routing, and this year (again) I forgot to do any optimization at all on the Dante VLAN and we never had a single network problem. 

But of course I believe every sound engineer should have some networking knowledge--in fact I wrote a book to address that issue :-)  But they also should (as you stated) know balanced transmission, soldering, etc etc.  But some very good mixers I know really aren't very technical.

Also I will say that in the audio and Dante training I do, the biggest misunderstandings I see are issues with understanding clock and latency, and of course both of those issues exist in all digital systems.  The advantage of networking is that once you get it, you get it, and it applies to many departments. 

• I reiterate the fact that if you give someone a system with an RJ45 on it, somewhere down the line that is going to end up plugged into a wider network and somebody needs to think about that.

Dante is designed to work on large networks, so if you follow their recommendations plugging into a larger network (not that I recommend that!) may not be catastrophic; it might not even be noticeable.  Most likely the IT department's access control list will shut off that interface, but even if it doesn't, link-local addresses won't pass through a router.  Of course if a Dante system finds a DHCP server on the larger network the addresses of the devices could change but even then if all your stuff is on the same subnet Dante will likely find it and reestablish the subscriptions.  I've seen it find stuff through some very bizarre setups.

But I would argue that this mis plugging issue is not limited to networks; working with students all day I know users will find a way to plug anything into anything wrong.  MADI's BNC, for example, is used for many applications, so it could easily be mis plugged into an SDI feed, old composite or RGBHV video, word clock, time code, etc etc.   And of course with (non standard but very common) moving lights with 3-pin XLR you can plug your moving light DMX feed into a mixer, or intercom, etc.  Labeling and education is key in all systems, of course.

• I fundamentally disagree that Dante is “simple”. It is “simple” to plug a mic into a powered speaker and make noise; that doesn’t make analogue audio systems or their design “simple”.

Nothing in modern audio systems is simple, so it's a reality we have to address. It's way easier to teach analog signal routing to students, and we start with that but that's no longer the world we live in.  Networks offer huge advantages to us (think about the reduction cost of a snake alone, plus the routing flexibility) so users need to educate themselves. 

• Thanks to Mic and Charles who made the really valid point that the best approach is to pick the format that integrates best with the particular desk you are dealing with

That was also the first thing I said in my response :-)

Gotta run, everyone have a good holiday (for those of us in the US).

John

-- 
++
www.controlgeek.net
www.johnhuntington.photography

[Rich Walsh]

On 22 Nov 2017, at 14:08, John Huntington <jch3ny...@gmail.com> wrote:

On 11/22/2017 6:26 AM, Rich Walsh wrote:

I’ve slightly lost track of which comments are directed at who, but:

• I can’t enforce that end users of a system I spec take a course to know how to use it.

Do you really think your users can step up and route MADI through/to a Digico (or CL5, or whatever) without some sort of training or assistance?  Training/education is a reality of every system today.  At least Audinate has a pretty well done course available online for free.  I recommend it for everyone because even if Dante goes away tomorrow (unlikely) networked audio is here to stay.  People need to learn it.

Again, I don’t disagree that people should learn how to use their tools. I just can’t enforce that. There’s a big difference between being taught how to use a mixing console – which is expected – and how to use a nebulous concept such as Dante – which isn't. Real people, really doing this and not going to college don’t, on the whole, seem to expect to have to know about more than how to use the actual physical device in front of them. I base that statement on 10 years of recruiting at the National Theatre, and a year trying to implement a program that made use of the monthly cross-department training days we used to have until someone worked out we could make tangible money on those days instead of investing in our staff…

I read your book 15 or more years ago, along with the The Art Of Digital Audio – I think I even bought copies for the NT and/or the Royal Court – but I don’t often encounter people who even really understand how MIDI works. How many people on this list have studied wordclock, for example?

You can provide training as part of the package for the current users. You can’t expect that future users – casuals, etc – will get, or even want, that training. In my experience, that’s less of a possible concern with MADI and a mixing console than with network audio.

• I disagree that there are misconceptions about Dante in this thread. If you can cite a specific statement about Dante that is wrong then I would like to be educated.

I would start with your first statement, "... Dante seems to require a knowledge of IP systems that is beyond most sound engineers at the moment, " ; also people subsequently talked about using DHCP or fixed IP addresses for Dante. Audinate now recommends in their training that you just use link-local addresses (automatically assigned without a DHCP server), and this means for 90% of systems (and 100% of small, stand-alone systems) if you use an unmanaged switch you really need to know very little about networking--it really is plug and play.  Even on larger networks you don't need to do much.  We run the Gravesend Inn on a managed network with four switches and 7 vlans and routing, and this year (again) I forgot to do any optimization at all on the Dante VLAN and we never had a single network problem. 

But of course I believe every sound engineer should have some networking knowledge--in fact I wrote a book to address that issue :-)  But they also should (as you stated) know balanced transmission, soldering, etc etc.  But some very good mixers I know really aren't very technical.

Also I will say that in the audio and Dante training I do, the biggest misunderstandings I see are issues with understanding clock and latency, and of course both of those issues exist in all digital systems.  The advantage of networking is that once you get it, you get it, and it applies to many departments. 

OK, I think I’m flogging a dead horse here then. Yes, yes, you can just plug it in and it does just work. But someone WILL go “ooh network” let me join that to this network and suddenly it doesn’t work – and it won’t be obvious to most people why not. I just think you need to consider that, and acknowledge that there IS a big difference between audio-only protocols and systems that use IP along with printers and Facebook. But, do you know, I’m bored of my voice on this now too.

• I reiterate the fact that if you give someone a system with an RJ45 on it, somewhere down the line that is going to end up plugged into a wider network and somebody needs to think about that.

Dante is designed to work on large networks, so if you follow their recommendations plugging into a larger network (not that I recommend that!) may not be catastrophic; it might not even be noticeable.  Most likely the IT department's access control list will shut off that interface, but even if it doesn't, link-local addresses won't pass through a router.  Of course if a Dante system finds a DHCP server on the larger network the addresses of the devices could change but even then if all your stuff is on the same subnet Dante will likely find it and reestablish the subscriptions.  I've seen it find stuff through some very bizarre setups.

But I would argue that this mis plugging issue is not limited to networks; working with students all day I know users will find a way to plug anything into anything wrong.  MADI's BNC, for example, is used for many applications, so it could easily be mis plugged into an SDI feed, old composite or RGBHV video, word clock, time code, etc etc.   And of course with (non standard but very common) moving lights with 3-pin XLR you can plug your moving light DMX feed into a mixer, or intercom, etc.  Labeling and education is key in all systems, of course.

I think it is different. To me it’s like the difference between the respect which people used to show to expensive rack-mounted PCs that you paid a specialist to build for you to run (shudder) SFX, and the Mac you get down the shops to run QLab. Maybe I’m too pessimistic, and should surround myself with diligent students?

• I fundamentally disagree that Dante is “simple”. It is “simple” to plug a mic into a powered speaker and make noise; that doesn’t make analogue audio systems or their design “simple”.

Nothing in modern audio systems is simple, so it's a reality we have to address. It's way easier to teach analog signal routing to students, and we start with that but that's no longer the world we live in.  Networks offer huge advantages to us (think about the reduction cost of a snake alone, plus the routing flexibility) so users need to educate themselves. 

• Thanks to Mic and Charles who made the really valid point that the best approach is to pick the format that integrates best with the particular desk you are dealing with

That was also the first thing I said in my response :-)

Sorry, I thought I’d read everything really carefully – it looked like you were just agreeing, not introducing a new thought.

I’m not sure anyone’s going to gain much more out of this thread now, so I’m going to abandon it myself and do some thinking.

Thanks to all for your input.

Rich

[Sam Kusnetz]

Hello all

I’d like to add only one small detail: there are consoles out there which do not have sample rate conversion available on MADI inputs. If you’re using one of them, you have to either use an all-96kHz workflow until you hit analogue outputs, or use a separate hardware SRC device to bridge the gap.

In either case, there are pitfalls in the form of increased cost, complexity, or processor load on the Mac that has to run everything at 96kHz. Running 96kHz audio is substantially more work for a computer than running 48kHz audio, and QLab users running at 96kHz have occasionally encountered intractable problems with sample dropouts. This happens rarely, but not never.

This is not a vote for or against anything, just a data point worth looking out for.

Best

Sam

—

Sam Kusnetz | Figure 53

s...@figure53.com

--
Contact support anytime: sup...@figure53.com
Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/73CFF191-210E-4B64-9330-A0ECD1F1ED64%40mac.com.

[Paul Gotch]

On 22/11/2017 15:10, Rich Walsh wrote:
> You can provide training as part of the package for the current users. You can’t expect that future users – casuals, etc – will get, or even want, that training. In my experience, that’s less of a possible concern with MADI and a mixing console than with network audio.

That is the same reasoning why a console might be locked down by a
designer such than an operator doesn't fiddle with the bits the designer
doesn't want them changing. The operator doesn't have to know how the
entire system works but also should not touch it if they don't know of
it works.

> OK, I think I’m flogging a dead horse here then. Yes, yes, you can just plug it in and it does just work. But someone WILL go “ooh network” let me join that to this network

And people plug their iPods in via dodgy cables and etc.

IMLE if you make it look like a dedicated bit of hardware with ethercon
connectors on it and paint it a funny colour then people leave it alone.
So far no one seems to have worked out you can take a £150 switch put it
in a rugged box with ethercon on it paint it purple and write 'Dante
Switch' on it in big letters and sell it for £2000.

> and suddenly it doesn’t work – and it won’t be obvious to most people why not.

Same goes for lighting these days. It's getting to the point where
fixtures only have ethercon and don't have conventional DMX at all.

This entire conversation reminds of the massively difficulties the sound
world had adjusting to having fewer faders than channels. Everyone seems
to have got use to that now.

I also once failed to get a west end sound designer to understand the
difference between the number of DSP processing channels a system had
the number of physical inputs it had. Historically one was entirely
bound to the other however they are now distinct and patchable quantities.

> I just think you need to consider that, and acknowledge that there IS a big difference between audio-only >protocols and systems that use IP along with printers and Facebook. But, do you know, I’m bored of my voice on >this now too.

Returning to lighting DMX512 is built on top of RS-485 which is just a
standard serial protocol. Dante being built on top of IP and Ethernet is
really no different.

>>> I reiterate the fact that if you give someone a system with an RJ45 on it, somewhere down the line that is going to end up plugged into a wider network and somebody needs to think about that.

The place to solve this is is network access control on the 'corporate'
network so it just doesn't work if you plug something else in to it.

-p

[Charles Coes Lists]

Dante v. Optocore - also about what ecosystem you are in... Controlling everything from a single patcher is wonderful - so in DIGICO land Optocore makes a ton of sense. Tracking patches down between the Digico, Dante controller and a DME or MRX makes ones head spin (Says the person who now defaults to feeding optocore into in orange box for dante distribution because it's also incredibly flexible for drive systems.)

Charles Coes

cco...@gmail.com

"Forms and rhythms in music are never changed without producing changes in the most important political forms and ways" - Plato

-- 
Contact support anytime: sup...@figure53.com
Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
--- 
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/F3D5F46F-A0E3-45F7-AAB5-B2A976314E7B%40mac.com.

[Rich Walsh]

On 22 Nov 2017, at 15:18, Sam Kusnetz <s...@figure53.com> wrote:

there are consoles out there which do not have sample rate conversion available on MADI inputs

Bifurcating this thread, as further research has me stumped on this one. I can’t find many consoles that do have SRC on multi-channel formats at all…

If you have an A&H dLive you have to run the console at 96kHz, but you can clock it to a master clock of 48kHz and use that to drive your QLab rig – Dante or MADI. There’s a form of SRC on the inputs – SRC to a common clock. A&H Dante can only be 48kHz though…

You can use a Focusrite RedNet D64R to SRC a Dante source to a MADI output – I think?

Yamaha have SRC on the Dante in of the RSio64-D. I don’t know what that means in practice.

That’s all I’ve come up with…

In terms of an elegant system design with a console at 96kHz, the best I can do is 48KHz Dante into A&H or 48kHz Dante converted to MADI at 96kHz into a DiGiCo. I can find MADI to MADI SRC, but not Dante to Dante – which would of course have to present to the network as if two separate devices.

I feel like I must be missing something here!

Rich

[micpool]

There is probably some reason why this wouldn't work, but for a Dante to Dante 32 ch SRC could you use a Yamaha RSio64-D with 2 Dante MY-16 

I'll ask Kieran at Audinate if he knows any method to do this.

Mic

[Sam Kusnetz]

Hello All

The Rednet D64R can indeed do both input and output sample rate conversion, so that's a great tool for handoff between Dante and MADI regardless of sample rate.

The Yamaha RMio64-D does basically the exact same thing, so it's another great choice.

I would strongly encourage anyone using a MADI-based 96kHz-only console with QLab to consider using one of these devices, and running their QLab system at 48kHz, feeding Dante to one of these, and using SRC to feed MADI at 96kHz to the console.

I'm sure Jon Weston could hear the difference, but the rest of us likely cannot. And anyway he's using an analogue Cadac...

Best

Sam

Sam Kusnetz | Figure 53

s...@figure53.com

--

Contact support anytime: sup...@figure53.com
Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.

To unsubscribe from this group and stop receiving emails from it, send an email to qlab+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/95dc9c92-0724-4681-b001-98de6a8a6f3a%40googlegroups.com.

[micpool]

The problem we haven't found a confirmed solution to yet is;  QLab at 48kHz Dante out to Console with Dante   at 96kHz.

Mic

On Wednesday, November 29, 2017 at 2:32:28 PM UTC, sam kusnetz wrote:

Hello All

The Rednet D64R can indeed do both input and output sample rate conversion, so that's a great tool for handoff between Dante and MADI regardless of sample rate.

The Yamaha RMio64-D does basically the exact same thing, so it's another great choice.

I would strongly encourage anyone using a MADI-based 96kHz-only console with QLab to consider using one of these devices, and running their QLab system at 48kHz, feeding Dante to one of these, and using SRC to feed MADI at 96kHz to the console.

[Sam Kusnetz]

Yeah...

Although there is a very expensive way to do it, of course... use two RedNet D64Rs!

Dante @ 48kHz > RedNet A > MADI @ 96kHz > RedNet B > Dante @ 96kHz

Bob's your uncle, assuming he's quite well off.

Sam

Sam Kusnetz | Figure 53

s...@figure53.com

--
Contact support anytime: sup...@figure53.com
Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+unsubscribe@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/14a90200-db38-4321-a31c-971dbfb98fae%40googlegroups.com.

[Elliot Sinclair]

Just to put my two cents into this. There are numerous shows running in the West End with the system and QLab running at 96kHz, that have no problems at all. 

As for hearing the difference, I know this pops up quite a lot. The main reason for running at 96kHz is to half the latency through the system, I can count on one hand people who have said it's because it sounds 'better', the rest want to cut down all those 'couple of milliseconds' through numerous pieces of equipment that equate to a lot of milliseconds. I'm not saying I endorse either just adding another a view on 96kHz. And even John Weston sometimes uses digital... just look at the FOH console on 'An American In Paris' in London!

On Wednesday, November 29, 2017 at 2:32:28 PM UTC, sam kusnetz wrote:

To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

[Rich Walsh]

That’s very interesting – so Autograph’s experience is that it’s OK? Are these Mac Pros or Mac minis? Are they running just audio or video as well? Are the files at 96kHz or are the Macs doing SRC on the fly?

Not as important, what protocol are they connected with?

I don’t have an informed opinion either way on whether 96kHz is worth the effort, but I need to figure out how to support it if it is expected…

Thanks.

Rich

[Justin Stasiw]

Agreed with Elliot-

We have a number of shows on Broadway and a number of national tours, all running MADI at 96Khz to assorted systems with no issues. Primarily with RME interfaces. We are all on Mac Mini's, the minis are SRCing (most of the content is 48K) on the fly, and we have not seen any MADI related issues. Some shows have lighter sound effects usage, some heavier. They are all only running audio, no video. 

--Justin Stasiw

[sam kusnetz]

Hello all

I was just gently teasing Mr. Weston; I have nothing but respect for him.

I’m very glad to hear confirmed reports of no problems with 96kHz audio and QLab.

To be clear, I don’t believe we’ve ever had enough data to be sure of anything, but only several examples where a system that should be performing perfectly well was having inexplicable issues, and 96kHz MADI and a console without SRC was a common thread.

Best

Sam

--

Sam Kusnetz | Figure 53

http://qlab.tips | http://qlab.tv

(mobile)

--
Contact support anytime: sup...@figure53.com
Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/0f725662-b4d2-4890-8726-3566c0b84513%40googlegroups.com.

[micpool]

I had a brief go at putting some numbers to running 48kHz files with SRC in QLab to 96kKHz, and inevitably it's difficult to come up with any definite conclusions

I was using a 10 ch USB interface with a 512 buffer in QLab

Interface set to 48kHz

Running 20 6 ch interleaved wavs.  (126 ch) sending to all 10 outputs

Audio good quality

CPU Load 22%  (only first thread on each of 4 cores used)

Interface set to 96kHz

Audio poor quality (distortion and dropouts)

CPU Load 30%  (only first thread on each of 4 cores used)

Doubling buffer to 1024 in QLab improves matters slightly but not usably.

Buffer back to 512 

Number of tracks halved 10 6ch interleaved wavs (60 tracks)

Audio Quality Good

CPU Load 22% (same as 126 tracks at 48kHz)

Presumably the data rate from the SSDs is the same regardless of interface sample rate. 126 tracks is around 17.3MB/s  i.e not very high. So it's not obvious where the bottleneck is.

For comparison 2 4K ProRes 422 videos playing simultaneously in QLab use 62% CPU (2 threads used on each core) VTDecoderXPCService using the bulk of that CPU. Data rate 134MB/s  Video Quality good. 

So my conclusion would be you can only run half the number of tracks if you are SRC 48kHz Files to 96kHz

I'll convert the files to  96kHz and do a comparison

Mic

[micpool]

With the files converted to 96kHz audio quality and usable  track counts similar with interface at 96kHz CPU about 2% lower with no SRC

Interface at 48Khz with SRC downsampling,  sound quality not as good as 48 kHz files 48kHz  interface in the previous test

So again nothing conclusive but 48kHz files/ 48kHz interface would appear to give maximum track counts with  good quality audio which isn't a surprise. 

48k files and 96k files to a 96k interface have similar performance.

However data rates and CPU usage still seems quite modest when audio starts falling apart. with 48Khz files SRC to Interface at 96kHz. and with 96kHz files to 96kHz interface. So unclear what the limiting factor is on track count.

It would be interesting to repeat the test with a  high quality 32ch interface.

A word of warning: The audio can get a bit crunchy with some bangs on some interfaces when things really start falling apart so you probably don't want to test this on show systems or expensive studio monitors!

Mic

[Tweaky]

Hello all,

Another two cents.

Since the beginning of this year I'm specifying a digital desk that doesn't offer SRC on the MADI inputs (or outputs). Some other desks do, by the way. Choosing this desk - for several other reasons - forced us to run the QLab files we have through an audio interface that clocks on 96K. It would have been possible to specify an expensive samplerate converter of course, but in the end there is always somebody who has to pay for it.

We run QLab mostly on the latest Mac Mini's in a Sonnet Mini Server enclosure with RME HDSPe MADI FX cards and preferably an extra SSD for media playback (via the Sonnet PCIe). Most of the files are being converted to 96K, some of them being multichannel (more than 4) audio files. Let's keep fingers crossed but I've never had problems with these setups. No distortion, hickups whatsoever. In one occasion the system also runs a heavy surround reverb/panner in Max, also at 96K.

Of course there will always be a discussion about the benefits or downsides of using a system at a certain samplerate, but I guess it all starts with the important question: what do you want to do? What are the technical prerequisites that are in play and can I (artistically) achieve what I'd want to. And once you have plan, stick to it and make it work the best you can. That sometimes means you have to cut down on channels (I had to in several occasions) or gain those extra milliseconds because you have to.

This forum and input from its members really help in achieving that, by the way... :-)

Cheers,
Dennis





Op vrijdag 1 december 2017 03:02:01 UTC+1 schreef sam kusnetz:

[Rich Walsh]

Could you indulge me with which desks please? I’ve not found them…

DiGiCo don’t*. The A&H dLive sort of does, for a common clock. I haven’t looked too deeply into Yamaha because the 16 channel slot architecture makes MADI an odd choice when the desk offers native Dante – but generally without SRC.

Big thanks to Mic Pool, by the way for taking the trouble to try to look at this scientifically. It reminds me of when Pro Tools had to have its own SCSI bus on the cards in order to get audio data through fast enough!

Rich

*It is possible to derive a half-speed split from an SD Rack, but not to receive an input on a different clock.

[Tweaky]

Hello Rich,

The SSL (L500) offers SRC on the Dante expander card.

To be quite honest: when writing I thought I was on the MADI inputs as well. But on the specific production I used the ability to SRC on the AES outputs to 'clock down' a System 6000 reverb, and not the MADI.

Dennis

Op vrijdag 1 december 2017 10:23:03 UTC+1 schreef Rich Walsh:

[Elliot Sinclair]

Ye all fine and working perfectly from an Autograph point of view. Majority are Mac mini systems in a Sonnet xMac Mini Server with an RME MADI cards and additional SSD drive. We do still have some old style Mac Pros that run fine as well, we don't own many of the new style Mac Pro dust bins so can't comment on them. All running audio only and I imagine a lot of it is SRC on the fly but I'll leave that for designers to confirm. 

We haven't got any Dante systems out at 96kHz at the moment to comment, I don't think...

The Yamaha Dante PCIe cards are very impressive and I think will be a big player in the future, they can support 128 channels at 96kHz but getting that out of a system is troublesome for us at the moment. They'd make great multi track recorders if we could get those channels out of a Digico system (DMI card is limited to 32 channels at 96kHz). Rather than having to use all of the MADI IO on the back an SD7 engine to do the same with multiple MGBs compared to a single PCI slot card.

The SD rack MADI split downsample to 48kHz is great, we use it for radio mic monitoring, especially when you've got more than 32 channels of RF on one SD rack. 

Thanks to Mic for taking the time to run those tests, makes for an interesting read.

The other point to throw into the mix for 96kHz MADI is frame rate, you can walk into limitations as we found out when a DD4MR only outputs SMUX MADI on a Digico Loop (it will support Hi Speed but not in a Digico system) and a DME64 only accepts Hi Speed MADI at 96kHz for example.

Elliot

[Rich Walsh]

That is not obvious from their website; in the end I had to use a Google site: search to finally discover http://livehelp.solidstatelogic.com/Help/SetupIO.html and:

If the Dante network is running at a different sample rate or in a different clock domain to the console, engage the SRC In button. This will sample rate convert incoming and outgoing audio between the console and Dante network. The number of channels available for routing on the console will vary based on the console and Dante network sample rates.

If you didn’t know you’d never had guessed!

Rich

[Maik Waschfeld]

Hi everyone,

Today a hint from Audinate:

<https://www.audinate.com/products/dante-enabled/studio-technologies/studio-technologies-model-5401-dante-master-clock>.

With kindest regards…
…Maik Waschfeld

(sent from my MBAir11)
also at <mailto:Maik.Wa...@Staatstheater-Stuttgart.de>

[Rich Walsh]

I think that’s just a high quality clock, like a Dante Big Ben.

Rich

[mackerr]

It is a high quality clock, but the important part is that you can sync that clock to video black burst. This is important in mixed audio/video environments like broadcast, where Studio Technologies sells most of their products. 

Mac

[Rich Walsh]

Sure: but what I was trying to find out in this thread was how people are managing more than one sample rate in their system – as advised by Sam – when using 96kHz desks and multichannel transport formats for which SRC is generally not available on the desk and so requires expensive third party kit – or, in the case of Dante, doesn’t seem to be possible at all except with some very specific setups.

Rich

[rob...@gmail.com]

Hi Rich, just to share our experience, we are using a dual redudant playback system in MADI, and are happy with it. 

This system is dedicated to a venue, so more than one operator have to use it.  Had to be easy and safe to operate. 

Console : Avid S6L-192, with one MADI card (64ch in x 64 ch Out) @ 96Khz, installed in the E6L engine.

The playback system consist of 2x Mac minis (quad core i7, Mac os 10.11) with SSDs, 2x Antelope audio Orion32 interfaces(32ch @96Khz), their MADI OUTs connected to a Directout Technologies MADI A/B switcher, connected to the Avid MADI card input 1.

The two Orion32 receives their clock from the E6L Engine, so they are also running at 96Khz.

The audio files in QLab are mostly at 44,1 or 48Khz, rarely at 96khz.

QLab plays between 16 to 24 channels of audio simultanously, rarely under 16, rarely over 24.

It is used for audio only and for some MIDI and OSC triggering to a Grand MA console and two Video Mac Pros running Millumin 2 or QLab 4, or a Coolux system if needed.

We installed the S6L last august, so I used the MADI playback system on about 40 shows only, but never had any trouble or any bugs.

So far so good.

It was the most logical solution for us because we already had the playback system, but were using the analog outputs from the Orion32. So we only needed to get the MADI AB switcher.

Dante was not an option with the S6L, unless Avid release a high density Dante card for the E6L Engine.

AVB will be an option in the next future, but right now there is no way for redundancy. Should be in the next S6L software update though.

Robert Caux

Grand Theatre de Quebec

[Rich Walsh]

Thanks: that’s very useful. Internal SSDs on the Mac minis, not any kind of fancy Sonnet box?

Rich

[so...@sorenknud.dk]

I think it would be good if there was a clearer indication of the samplerate of your files in Qlab, and if weather they match your hardware sample rate.

BTW: Sorry off topic, but Robert are you happy with the S6L ?

Best

Soren

[Chris Ashworth]

Hi Soren,

This might be a silly question of me, but: why? Generally speaking it doesn’t matter.

-C

[Andrew Keister]

Hi Chris,

While I agree with you that frequently it's not important information, I think there are use cases when it is important to be able to minimize the processor loads as much as possible. I'm generally well behaved when it comes to outputting my files for a project in the correct format, but I've walked into projects with all kinds of compressed formats and mismatched sample-rates being used in manners that were causing some performance issues. It's not that hard to find the target in the OS and get info there, but it's not elegant either. I've alway thought it would be nice to have file information about the target either in the basics tab or it's own location simply for convenience.

Best,

Andrew

[Rich Walsh]

I could tart it up a bit, but you could put this on a hot key:

tell application id "com.figure53.QLab.4" to tell front workspace to display dialog "Sample rate of selected cue: " & (do shell script "mdls " & quoted form of POSIX path of (file target of last item of (selected as list) as alias) & " | grep 'kMDItemAudioSampleRate' | awk '{print $3}'")

Rich

-- 
Contact support anytime: sup...@figure53.com
Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
--- 
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/45cee7a0-dc45-46c9-88b7-bd7e2a2302e3%40googlegroups.com.

[so...@sorenknud.dk]

Well possibly its just my OCD ;)

I like the way an app like Millumin can tell me if I have "not optimal" files (they will play without problems, but I can go into an "optimize" menu and check if have forgotten to convert a file). Not an "error message" just a volentary "this other fileformat could possibly make it easier on your computer".

Not a important feature at all, but nice to have someday.

Best regards,

Soren

[micpool]

If you want to scan through a lot of cues quickly to find sample rate outliers then you are probably better off having a list rather than having to look in the inspector of every cue.

Here's a script using Rich's suggestion of  mdls, (I'd previously been using afinfo in shell scripts e.g. set shelltext to "afinfo " & fr & " | grep \"bit rate\""  but this is better because everything is separated.)

If you are going to use this on a hotkey make it obscure as it puts up file dialogs. I'm using ⇧⌃⌥R

As a novelty It will show it's progress in the  Active Cues pane. (This probably slows it down considerably!)

When it has compiled a list of files, sample rates, and bit depths, it will ask you for a filename and then save as a text file. You can either read this directly or import it into excel using comma separators.

As well as finding inconsistent sample rates it's also good for checking file paths to make sure everything is coming from the folder you think it is.

I am sure it can be improved....

Here's the script

tell application id "com.figure53.QLab.4" to tell front workspace

set mycue to last item of (active cues as list)

set the q name of mycue to "Produce file of sample rates"

set foundcues to (every cue whose q type is "audio")

set filelist to a reference to foundcues

set outputData to {}

set thecount to (count of filelist)

set progrestotalsteps to thecount

set progresscompletedsteps to 0

set filesprocessed to 0

repeat with thisfile in filelist

set thetarget to (file target of thisfile) as alias

set thepath to quoted form of POSIX path of thetarget

set shelltext to "mdls " & thepath & " | grep 'kMDItemAudioSampleRate'  "

set theSampleRate to do shell script shelltext

set shelltext to "mdls " & thepath & " | grep 'kMDItemBitsPerSample'  "

set theSampleBit to do shell script shelltext

set textOutput to thepath & "," & word 3 of theSampleRate & "," & word 3 of theSampleBit

set the end of outputData to textOutput

set filesprocessed to filesprocessed + 1

set progresscompletedsteps to filesprocessed

set the q name of mycue to "Processing " & progresscompletedsteps & " of " & thecount

end repeat

set the q name of mycue to "Save file of sample rates"

set AppleScript's text item delimiters to return

set outputData to outputData as text

set AppleScript's text item delimiters to " "

set theNewFilePath to choose file name with prompt "Save a text document with Sample Rates Of All Files:"

set filereference to open for access theNewFilePath with write permission

write outputData to filereference as text

close access filereference

set the q name of mycue to "Produce file of sample rates"

end tell

An example bundle is attached 

Mic

[Rich Walsh]

Here are some quick optimisations for you:

tell application id "com.figure53.QLab.4" to tell front workspace

set mycue to last item of (active cues as list)

set the q name of mycue to "Produce file of sample rates"

set foundcues to file target of cues whose q type is "audio" -- ### Reduce the number of calls by getting all of these in one hit

set filelist to a reference to foundcues

set outputData to {}

set thecount to (count of filelist)

-- set progrestotalsteps to thecount ### Unused variable

set progresscompletedsteps to 0

set filesprocessed to 0

repeat with thisfile in filelist

-- set thetarget to (file target of thisfile) as alias ### Not iterative (see above)

-- set thepath to quoted form of POSIX path of thetarget ### Variable change, and also save quoted form for shell script to avoid ' in text file

set thepath to POSIX path of thisfile -- ### See above

(* set shelltext to "mdls " & thepath & " | grep 'kMDItemAudioSampleRate'  "

set theSampleRate to do shell script shelltext

set shelltext to "mdls " & thepath & " | grep 'kMDItemBitsPerSample'  "

set theSampleBit to do shell script shelltext

set textOutput to thepath & "," & word 3 of theSampleRate & "," & word 3 of theSampleBit *)

tell me to set sampleInfo to do shell script "mdls " & quoted form of thepath & " | grep -e kMDItemAudioSampleRate -e kMDItemBitsPerSample" -- ### Tell me to save -10004 error as QLab can't run shell scripts inside a tell block; get both values in one hit to save overhead

set textOutput to thepath & "," & last word of paragraph 1 of sampleInfo & "," & last word of paragraph 2 of sampleInfo -- ### Changed for two values in one variable

set the end of outputData to textOutput

set filesprocessed to filesprocessed + 1

set progresscompletedsteps to filesprocessed -- ### Why not just use filesprocessed in line below and cut this variable?

set the q name of mycue to "Processing " & progresscompletedsteps & " of " & thecount

end repeat

set the q name of mycue to "Save file of sample rates"

set AppleScript's text item delimiters to return

set outputData to outputData as text

set AppleScript's text item delimiters to " "

set theNewFilePath to choose file name with prompt "Save a text document with Sample Rates Of All Files:"

set filereference to open for access theNewFilePath with write permission

write outputData to filereference as text

close access filereference

set the q name of mycue to "Produce file of sample rates"

end tell

I have a very old routine called “Make a list of media files” that I was thinking about updating on the next template refresh to include this info. I used to have a really cool table in QLab Script Companion with the Audio Toolkit for SRC, convert to wav, interleave, append silence for looping, etc – but never managed to maintain that app when Xcode changed massively and I had to relearn how to link UI objects to the script. Still haven’t, 8 or so years later!

Rich

--
Contact support anytime: sup...@figure53.com
Follow Figure 53 on Twitter: https://twitter.com/Figure53
User Group Code of Conduct: https://figure53.com/help/code-of-conduct/
---
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/8a171349-1cb2-43c2-909d-f0bc3a5e8d18%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

<File list.zip>

[micpool]

Thanks, Rich that's much faster.

There were a couple of audio files missing in the previous demo.

So attached is a new bundle with Rich's optimisations in the script and all the audio files