Issues going through a Mac for on-line jams?

[ssjk]

I’ve been trying with mixed results to use JamKazam with some friends who are around 300 miles from me. First tried with an old AudioBox iTwo and had a mediocre 11.4 ms latency between me and the internet. Not good enough.

So I upgraded to a gen 3 Focusrite Scarlett 212. Jamkazam showed the same 11.4 ms lag for the Scarlett when first hooked up, but Focusrite has a new MacOS patch that drops that down to a more usable but still kind of sketchy 6.4ms.

For comparison, both of my friends running PC/Windows boxes show a not quite believable 0ms (as in Zero) gear latency. My keen analytical skills point to the MacOS/Apple hardware combination as a potential issue. Maybe because the PCs have separate sound processors and the MacBook doesn’t?

Has anybody been able to do remote jams using a Mac as the link between the mic and the internet? If not, does anybody know if this i just the nature of the beast with Apple - no way to get around that interface lag.

Kind of need to figure this out before I order that Mac mini I have been planning on.

Thanks

[FrankHudson]

Originally Posted by ssjk

I’ve been trying with mixed results to use JamKazam with some friends who are around 300 miles from me. First tried with an old AudioBox iTwo and had a mediocre 11.4 ms latency between me and the internet. Not good enough.

So I upgraded to a gen 3 Focusrite Scarlett 212. Jamkazam showed the same 11.4 ms lag for the Scarlett when first hooked up, but Focusrite has a new MacOS patch that drops that down to a more usable but still kind of sketchy 6.4ms.

For comparison, both of my friends running PC/Windows boxes show a not quite believable 0ms (as in Zero) gear latency. My keen analytical skills point to the MacOS/Apple hardware combination as a potential issue. Maybe because the PCs have separate sound processors and the MacBook doesn’t?

Has anybody been able to do remote jams using a Mac as the link between the mic and the internet? If not, does anybody know if this i just the nature of the beast with Apple - no way to get around that interface lag.

Kind of need to figure this out before I order that Mac mini I have been planning on.

Thanks

I'm replying even though I'm puzzled by the your post in some ways and not some kind of thorough-going expert on these matters.

There's no substantial difference in the supplied sound-processors in typical Mac computers and typical "Windows" computers, but even if there were, your Focusrite interface is now (or should be) your sound processor. When recording on a computer, besides the step of converting your just created sound waves of your instrument or voice to digital information you may choose to also modify that sound (for example with a reverb effect or a software based compressor running on your computer). The former is pretty easily handled by any modern computer, the later can have various complexity and computational loads and complexity.

Many/most sound interfaces allow "zero latency" by way of a setting or a knob that lets you skip any processing being done by your computer and supplying instead what you are playing into the interface coming back right back to you without the effects. In the olden days this was more necessary more often than it is with modern computers when using effects in a moderate way.

The Internet on the other hand has tremendous latency compared to what you do on your computer. This is why trying to do a real-time chat even with voice and video is prone to funny out of sync effects, like a bad dubbing job on an old foreign movie.

It's common for people to confuse latency and bandwidth with computer networks/connections. The example I would often use when describing this difference is that if I was to fill up the back of large truck with boxes of filed large storage capacity hard drives and have it sent across town that would have tremendous bandwidth, but also considerable latency issues. If one of those hard drives had a yes/no answer, or a small sound file you need right now you wouldn't want to wait for the truck to get across town, get unloaded and for your important small bit of data to get retrieved.

I have no experience with the application you're trying to use to jam online or how it works to overcome this, and how it can be optimized. Sorry. In my experience, even with older computers (my main recording Mac which is hooked up to a 16 channel interface is a 2011 model) local recording can work well on Macs.

[ljguitar]

Hey ssjk

Here's my suggestion…

Go to the Acoustic Amplification sub-forum, and lookup the post started by dcopper titled:

Internet Practice Platform? posted on 4/16/2020

Read the entire thread, because Doug Young starts talking about JamKazam in vague terms, which turned into full-on Doug Young experimentation, with statistics over the next 20 posts (with charts and stats).

If that's not enough to start with, he'd be the guy I'd private message, and pick his brains about it.

Hope this helps…

[ssjk]

Larry - good suggestion. I read that thread a while back and did not recall any conversation about mac vs pc, but will go back and see.

Frank - Thanks for the response. I understand latency (lag) vs capacity and the issue here is definitely the former. I also know by reading and direct experience that above 20-25 ms total the latency makes playing together near impossible. Our measured internet latency for these sessions is 11.2ms, so it’s really a question of latency associated with our interfaces that is the issue.

I look at it this way: the signal on a one-way trip between my guitar and the ears of my playing partner has to go through these steps:

1 Analog to digital conversion (by the interface)
2 grab it and pass it along (by the computer)
3 run it down the internet (depends on distance and other things)
4 grab it and pass it along (other player’s computer)
5 digital back to analog conversion (by their interface)

the combination of 1+2 in my case using a mac for #2 was 11.4ms until I did the patch provided by Focusrite for MacOS, at which point it went down to 6.4ms.

And the equivalent latency for my friends running PC/Windows for their item 2 registers as zero. One is running the same Focusrite interface, one has a newer version of the PreSonus interface.

Clearly there is something the mac is doing in the grab and pass along stage that takes 5ms less time than it did before the patch, and that still takes 6.4ms longer than it does on my friends side. I really just want to know if this is a systemic mac issue before I buy another one.

For the PreSonus Audiobox i2 and the Focusrite scarlett 212 I am unable to find any software that allows me to adjust sample size or anything else. Nor can I find anything in the Mac sound preferences to do that. I’d love it if someone could point me to the correct place to do that.

By the way, either interface works just fine for recording, and I can adjust for direct monitoring or with processing at the interface for that purpose. That seems to be moot for these purposes.

[Chipotle]

I am curious how you are measuring latencies. Is it something JamKazam is telling you, or something else? JamKazam could only tell you its own (+internet) latency between the two users. I'm not sure how it would know what the audio system on your end is doing.

PCs don't have "separate sound processors" when you're going through a USB interface. The application just sends the data to the USB port and lets the interface do its thing.

I have a 2i2 as well, and can pull up Focusrite Device Settings that has a buffer size, which does affect latency. You want as small a buffer as you can without getting cracks/pops or other audio artifacts.

[ssjk]

Originally Posted by Chipotle

I am curious how you are measuring latencies. Is it something JamKazam is telling you, or something else? JamKazam could only tell you its own (+internet) latency between the two users. I'm not sure how it would know what the audio system on your end is doing.

>> I think JamKazam tests the interface and measures it. It can tell whether the connection is wired or wireless and reports the latency figures along with other data like jitter.

PCs don't have "separate sound processors" when you're going through a USB interface. The application just sends the data to the USB port and lets the interface do its thing.

I have a 2i2 as well, and can pull up Focusrite Device Settings that has a buffer size, which does affect latency. You want as small a buffer as you can without getting cracks/pops or other audio artifacts.

>> "PCs don't have" may be the heart of the issue. I infer you are on a PC and can do change buffer size and other settings through s focusrite management app. I cannot do this on the mac - no such place to do device setting through focusrite or through the mac sound preferences. If somebody knows where they are please tell me!

[KevWind]

I don't know the answer
But I have some thoughts to perhaps consider ..

#1 there is no such thing as (0) Zero latency when involving A/D and D/A conversion, it is usually not detectable but is there in all converters

# 2 it is possible but not likely, there is any significant "measurable" or noticeable difference in the digital throughput between a Mac and a PC

#3 when using an interface the input A/D conversion is happening in the interface not the computer same with the output D/A conversion

#4 it could be possible the the Mac OS is doing something with routing causing additional latency (but again not likely) or that the Focusrite drivers for Mac are adding latency that is not in it's Windows drivers?

#5 If you saying the JamKazam website is reporting the latency then your
specific ISP router and internet upload/download speed is definitely a factor also .

So it sounds like you are going from your Interface directly into the JamKazam App ? and not going through a DAW first ? So if ( buffer sample rate) it is not a setting in JamKazam preferences it is possible that you have no buffer settings you can adjust like you can in most DAWs I am presuming you or JamKazam) are using 48 k as the sample rate ..

[Chipotle]

Originally Posted by ssjk

>> "PCs don't have" may be the heart of the issue. I infer you are on a PC and can do change buffer size and other settings through s focusrite management app. I cannot do this on the mac - no such place to do device setting through focusrite or through the mac sound preferences. If somebody knows where they are please tell me!

I guess it's because Apple uses its own audio drivers, whereas on Windows you get to use Focusrite-supplied software. Without going into any Apple-bashing here I did find this. Not sure how much help it might be:

https://support.focusrite.com/hc/en-...erfaces-on-Mac

[ssjk]

I did contact Doug Young, who was kind enough to reply and confirm my suspicions:

" From what I understand, everything on a Mac goes thru Core Audio, which indeed adds about 6ms latency, and there's apparently not much you can do about it."

PCs you may or may not have better results and/or ability to set some parameters.

Sometimes you love Apple's lockdown on user meddling, sometimes you hate it. It's that kind of world. I shall persevere.

[Doug Young]

For anyone trying JamKazam, I'd suggest checking out this document:

https://docs.google.com/document/d/1...mf8dYv82Z6irQI

Lots of info, including how to minimize latency to the extent you can. JamKazam is also helpful at showing what's going on. I posted one screen in the Internet Practice thread. Here's another example of the data you get:

stats.jpg

Lots of data that's hard to interpret, but basically you want to see lots of green and no red :-)

The hard part is doing something about this, especially the internet part. But you can try to minimize what you can, by not using wifi and doing some router configuration (discussed in the above doc).

One thing about latency to keep in mind is the 6ms for the mac, while frustrating, is actually pretty low. To put it in context, imagine you are using a mic and the mic is 2 feet away from your guitar. Sound travels 1 foot/millisecond, so you have 2ms lost to the "air", alone! You can also lose more to your interface, depending on settings. For example, my Apollo has some built-in delay compensation that is useful when overdubbing, etc. When using JamKazam, I turn that off. Of course it's kind of beside the point, if the internet is taking 100ms....

[KevWind]

This thread caused me to do some more research On ASIO vs Core Audio and as usual there is a mass of info. And some of gets pretty complicated and
Unfortunately much of the linked info there is conflicting info as well. And a fair amount of debate

But from what I gathered in a brief search:

Mac machines use Core Audio (which is Apple hardware dependent) and which comes with the OS. Windows machines can use several different ones, but ASIO (Which is hardware independent ) is the most popular with the Audio interface manufactures .
Apparently there are a number of variables involved and so a lot of debate as to whether (all things considered) ASIO or Core Audio might be faster (lower in latency) and or more accurate and user friendly

One answer I read that made sense to me (but I do not know if actual measurements back it ) Is that ASIO makes less "Kernel calls" per sample, than Core Audio, making it a bit more efficient ?
One advantage to Core Audio is it is said to be multi-client capable, meaning, not only can it run your DAW, but all your other computer and internet sound through your audio interface and recording sound system as well.

I stumbled on one list of comparison Ms latency posted on Gear Slutz by a member (so take that as it is ) , based on different buffer sample sizes and it did show that ASIO was reporting and average of 1-2 ms faster, round trip times than Core Audio . ?????

Doug I am curious where did you get the 6 ms number for Core Audio ?

[Doug Young]

Originally Posted by KevWind

Doug I am curious where did you get the 6 ms number for Core Audio ?

I haven't researched it myself, that was just the reported number from JamKazam for my audio interface. The guys who wrote that paper I posted told me that with CoreAudio that was the best I was going to get with a Mac, but that it could be tuned lower with a PC, but not with a Mac. It's not clear to me that JamKazam has any way to measure my real total latency (i.e from the mic, thru my preamp/interface, etc), so I assume its reporting mostly something it can detect about the audio drivers (maybe a loopback time). That said, I've worked with others on Macs, with other interfaces, who were much slower (as reported by JamKazam), so I'm not sure what's included in the measurement.

For what it's worth, I've never seen an interface latency reported by JamKazam of less than 4ms, even with people with PCs - might be better out there, but I have yet to see it, and often it's much more. So, I think we're really talking about a smaller difference between Mac and PCs than 6ms.

Generally, if the internet's ok, it works. I've had reasonable sessions with latencies JamKazam labels as poor, and when the internet connection is good, it works very well, so shaving off a few more ms hasn't been a priority. I'd say the biggest annoyance so far has been that the internet is variable. One second it's fine, and the next, latency skyrockets, dropouts start happening, and so on. Getting off wifi and setting up my router config did seem to help a bit.